v1.5.1 "A2A Governance" — Now Available

Enterprise Governance for
Autonomous AI Agents

AgentShroud sits transparently between your AI agents and every system they touch. Every tool call, API request, file write, and cloud mutation is intercepted, inspected, and policy-enforced — without modifying the agent.

# every request through every security module, every time
AI Agent  →  AgentShroud Gateway  →  Target System
            └── 75 modules │ 7 layers │ <0.5ms P99
75
Active security modules — zero stubs
4,300+
Tests — 85% coverage
<0.5 ms
Inbound latency P99 (arm64/macOS)

Production-grade from day one

Every module is fully wired in the request pipeline. No dead code, no placeholders.

🛡️

Prompt Injection Defense

PromptGuard + HeuristicClassifier intercept and neutralize injection attempts before they reach tool execution.

🔍

PII Sanitization

Presidio-powered detection at 0.9 confidence minimum. Redacts PII from both inbound and outbound traffic.

🌐

Egress Filtering

Domain allowlisting, DNS filtering, and real-time egress approval queue with Telegram notifications.

✋

Human-in-the-Loop

High-risk actions — email sending, file deletion, external API calls, skill installation — route through a persistent approval queue.

👥

SOC Collaboration Dashboard

7-page web control center: approvals, delegations, tool ACLs, shared memory, RBAC, and live agent activity.

📊

Daily CVE Reports

Trivy-powered CVE digests delivered daily via Telegram. Severity-filtered, owner-gated, with error-report fallback.

🔐

Credential Hygiene

1Password CLI, macOS Keychain, and secret-tool integration. No plaintext secrets on host disk during normal operation.

📦

Supply Chain Integrity

Cosign image signing, Trivy CVE scanning, Semgrep SAST, ClamAV malware scanning, Falco runtime enforcement.

🎯

Adversarial Red-Team Assessment

STPA-Sec methodology (Nancy Leveson, MIT), developed with security advisor Steven Hay — code-path and live assessments run through the production interface, not a synthetic benchmark.

⚖️

IEC 62443 Alignment

FR3→SL3 system integrity, FR6→SL3 audit logging with SHA-256 hash chains, FR7→SL2 availability hardening.

7 defense layers, 75 modules

Every inbound request traverses all active layers. Enforcement is the default — monitor mode requires explicit opt-in.

Layer Modules
P0 Core PromptGuard, TrustManager, EgressFilter, PII Sanitizer, Gateway Binding
P1 Middleware SessionManager, TokenValidator, ConsentFramework, SubagentMonitor, AgentRegistry, Delegation, SharedMemory, ToolACL, PrivacyPolicy, RBAC, ContextGuard, ContextIntegrity, EncodingDetector, InputNormalizer, HeuristicClassifier, ApprovalHardening, BrowserSecurity
P2 Output OutboundFilter, OutputCanary, OutputSchema, MultiTurnTracker, LogSanitizer, MetadataGuard, ToolResultSanitizer, ToolResultInjection, XmlLeakFilter, PromptProtection
P3 Tool Control ToolACL, ToolChainAnalyzer, InstructionEnvelope, SubagentMonitor, ProgressiveLockdown, KillswitchMonitor
P4 Network EgressFilter, EgressApproval, EgressMonitor, DNSFilter, NetworkValidator, WebContentScanner
P5 File & Memory FileSandbox, PathIsolation, DriftDetector, MemoryIntegrity, MemoryLifecycle, GitGuard, EnvGuard, ConfigIntegrity
P6 Infrastructure ImageVerifier (cosign/Sigstore), ClamAV, Trivy, Falco, Wazuh, EncryptedStore, KeyVault, KeyRotation, HealthReport, AlertDispatcher, CanaryTripwire, SOCCorrelation, ScannerIntegration

OpenClaw

OpenClaw is AgentShroud's flagship production bot — a Node.js autonomous agent shipped as agentshroud-openclaw. It has been in production since v1.0.x and is the primary interface for Telegram, Slack, iOS Shortcuts, Browser Extension, SSH, and REST API — all traffic governed by the full 75-module security pipeline.

💬

Multi-Channel Presence

Telegram bot, Slack workspace integration, iOS Shortcuts, Browser Extension, SSH terminal access, and REST API — every channel routes through AgentShroud before reaching any downstream system.

🔒

Security Posture

All egress routes via gateway:8181 (HTTP CONNECT proxy). All LLM calls route via gateway:8080. PII is redacted at 0.9 confidence before any response reaches a user. Credentials are never leaked in chat — blocked at the sanitizer layer.

🧠

Persistent Memory

Per-collaborator workspace isolation with encrypted memory volumes. Conversation context, task history, and collaborator profiles persist across restarts on the agentshroud-config volume.

🖥️

Web Control UI

Built-in web dashboard at port 18789 (published at 127.0.0.1:18789). Manage devices, view conversation history, and configure channels without CLI access.

Connect via Telegram
# Add the bot to your Telegram workspace
# Bot token stored in Docker secret: telegram_bot_token

# Check status
curl http://localhost:8080/status

# Full setup guide: docs/setup-telegram.md

Hermes Agent

Hermes is the second AgentShroud-secured autonomous AI agent — Python-based, built on nousresearch/hermes-agent. It runs alongside OpenClaw inside the same 75-module governance envelope, fully isolated on the agentshroud-isolated network with all LLM and egress traffic routed through the gateway.

🤖

Capabilities

Telegram, Slack, Brave web search, web fetch, cron scheduling, subagent personas, and native MCP (Model Context Protocol) server integrations — all policy-enforced at the gateway.

🔒

Security Posture

All Hermes egress routes via gateway:8181 (HTTP CONNECT proxy). All LLM calls route via gateway:8080. Same 75-module pipeline as OpenClaw — no bypass path exists at the container level.

🌐

OpenAI-Compatible API

Exposes an OpenAI-compatible API on port 8642. Any tool or client that speaks the OpenAI chat completions format can connect to Hermes through the gateway's credential-injection and audit layer.

🖥️

HCI Admin UI

Built-in admin dashboard on port 9119. Access is gateway-auth-gated — unauthenticated requests are blocked before reaching the Hermes process.

Connect to Hermes API — base URL
# OpenAI-compatible endpoint (requires gateway auth token)
base_url = "http://localhost:8642/v1"

# Health check
curl http://localhost:8642/health

# Full setup guide: docs/setup/HERMES_SETUP.md

Voice Terminal

The ESP32-S3-BOX-3 is an optional hardware control surface that lets you interact with any AgentShroud-proxied agent by voice — say "Hi, ESP" or tap the screen, speak naturally, and hear the agent's reply through the built-in speaker. All audio routes through the full 75-module security pipeline: PII redaction, prompt-guard, audit hash-chain, egress policy.

🎤

Wake Word + Tap-to-Talk

Say "Hi, ESP" or tap anywhere on the touchscreen. Short tap → VAD timeout (8 s silence ends utterance). Long press → ends on release.

🔄

Runtime Agent Toggle

Press the MUTE button to cycle between Hermes (agentic, synchronous reply), Fast LLM (low-latency Claude Haiku proxy), and OpenClaw (async — replies on Telegram).

🔒

Full Security Pipeline

Utterances route through POST /forward?route_to=<agent> — the same AgentShroud pipeline used by all other control surfaces. Token-authenticated WebSocket over Tailscale Funnel (TLS 443).

📡

Works Anywhere

Tailscale Funnel provides a public TLS endpoint on marvin. The device works on home WiFi, cellular hotspot, or any internet connection — no VPN client on the ESP required.

# Start the voice gateway
docker-compose -f docker/docker-compose.yml -p agentshroud --profile voice up -d

# Flash the firmware (ESP32-S3-BOX-3 on USB)
cd firmware/voice-terminal
idf.py -p /dev/cu.usbmodem* flash

# Full setup guide:
docs/integrations/voice-terminal-esp32-s3.md

891 OpenClaw CVEs — all mitigated

All CVEs disclosed in OpenClaw (January–September 2026) are documented and tracked by AgentShroud’s defense-in-depth architecture.

CVE CVSS Title Status
ASH-OCLAW-006 9.9 CRITICAL WebSocket Scope Self-Declaration ✓ Mitigated
ASH-OCLAW-009 9.9 CRITICAL Token Scope Expansion via device.token.rotate ✓ Mitigated
ASH-OCLAW-018 9.9 CRITICAL In OpenClaw before 2026.2.23, tools.exec.safeBins validation for sort could be bypassed... ✓ Mitigated
ASH-OCLAW-019 9.9 CRITICAL Versions prior to 2026.2.14 contain a vulnerability in the gateway in which it fails to... ✓ Mitigated
ASH-OCLAW-020 9.9 CRITICAL Before 2026.3.28 contains a privilege escalation vulnerability in the /pair approve com... ✓ Mitigated
ASH-OCLAW-021 9.8 CRITICAL CVE-2026-27002 ✓ Mitigated
ASH-OCLAW-022 9.8 CRITICAL Versions prior to 2026.2.2 fail to properly validate Windows cmd.exe metacharacters in ... ✓ Mitigated
ASH-OCLAW-023 9.8 CRITICAL Versions prior to 2026.2.2 contain an exec approvals (must be enabled) allowlist bypass... ✓ Mitigated
ASH-OCLAW-024 9.8 CRITICAL OpenClaw's Nextcloud Talk plugin versions prior to 2026.2.6 accept equality matching on... ✓ Mitigated
ASH-OCLAW-025 9.8 CRITICAL A remote code execution (RCE) vulnerability in OpenClaw Agent Platform v2026.2.6 allows... ✓ Mitigated
ASH-OCLAW-026 9.8 CRITICAL Before 2026.2.24 contains a sandbox network isolation bypass vulnerability that allows ... ✓ Mitigated
ASH-OCLAW-027 9.8 CRITICAL Before 2026.3.12 contains an authorization bypass vulnerability where Feishu reaction e... ✓ Mitigated
ASH-OCLAW-028 9.8 CRITICAL Before 2026.3.11 contains an exec allowlist bypass vulnerability where matchesExecAllow... ✓ Mitigated
ASH-OCLAW-029 9.8 CRITICAL Before 2026.3.12 contains a weak authorization vulnerability in Zalouser allowlist mode... ✓ Mitigated
ASH-OCLAW-030 9.8 CRITICAL Before 2026.3.13 allows bootstrap setup codes to be replayed during device pairing veri... ✓ Mitigated
ASH-OCLAW-031 9.8 CRITICAL Before 2026.3.13 contains a remote command injection vulnerability in the iMessage atta... ✓ Mitigated
ASH-OCLAW-032 9.4 CRITICAL Versions prior to 2026.2.1 with the voice-call extension installed and enabled contain ... ✓ Mitigated
ASH-OCLAW-033 9.4 CRITICAL Versions 2026.3.7 before 2026.3.11 contain an authorization bypass vulnerability where ... ✓ Mitigated
ASH-OCLAW-034 9.3 CRITICAL Before 2026.3.7 contains an improper header validation vulnerability in fetchWithSsrFGu... ✓ Mitigated
ASH-OCLAW-017 8.8 HIGH Privilege Escalation (operator.write reaches owner-only surfaces) ✓ Mitigated
ASH-OCLAW-035 8.8 HIGH (aka clawdbot or Moltbot) before 2026.1.29 obtains a gatewayUrl value from a query stri... ✓ Mitigated
ASH-OCLAW-036 8.8 HIGH (formerly Clawdbot) is a personal AI assistant you run on your own devices ✓ Mitigated
ASH-OCLAW-037 8.8 HIGH CVE-2026-26323 ✓ Mitigated
ASH-OCLAW-038 8.8 HIGH Versions prior to 2026.2.14 contain a command hijacking vulnerability that allows attac... ✓ Mitigated
ASH-OCLAW-039 8.8 HIGH Version 2026.2.22-2 prior to 2026.2.23 tools.exec.safeBins validation for sort command ... ✓ Mitigated
ASH-OCLAW-040 8.8 HIGH Versions prior to 2026.2.14 contain a path traversal vulnerability in apply_patch that ... ✓ Mitigated
ASH-OCLAW-041 8.8 HIGH Versions prior to 2026.2.25 contain a symlink traversal vulnerability in the agents.fil... ✓ Mitigated
ASH-OCLAW-042 8.8 HIGH Versions 2026.2.22 prior to 2026.2.25 contain a privilege escalation vulnerability allo... ✓ Mitigated
ASH-OCLAW-043 8.8 HIGH Before 2026.3.12 contains an insufficient access control vulnerability in the /config a... ✓ Mitigated
ASH-OCLAW-044 8.8 HIGH Before 2026.3.11 contains a sandbox boundary bypass vulnerability allowing leaf subagen... ✓ Mitigated
ASH-OCLAW-045 8.8 HIGH Before 2026.3.11 contains an authorization bypass vulnerability in the gateway agent RP... ✓ Mitigated
ASH-OCLAW-046 8.8 HIGH Before 2026.3.22 contains a privilege escalation vulnerability in the Control UI that a... ✓ Mitigated
ASH-OCLAW-047 8.8 HIGH Before 2026.3.22 contains a privilege escalation vulnerability in the device.pair.appro... ✓ Mitigated
ASH-OCLAW-048 8.8 HIGH Before 2026.3.22 contains an unvalidated WebView JavascriptInterface vulnerability allo... ✓ Mitigated
ASH-OCLAW-049 8.8 HIGH Before 2026.3.25 contains a privilege escalation vulnerability allowing non-admin opera... ✓ Mitigated
ASH-OCLAW-050 8.8 HIGH Before 2026.3.22 contains an allowlist bypass vulnerability in system.run approvals tha... ✓ Mitigated
ASH-OCLAW-817 8.8 HIGH Google Meet node commands could skip exec approvals ✓ Mitigated
ASH-OCLAW-827 8.8 HIGH Slack group DMs could skip sender allowlists ✓ Mitigated
ASH-OCLAW-844 8.8 HIGH Cron tool could accept mixed-case command payloads ✓ Mitigated
ASH-OCLAW-847 8.8 HIGH Codex bind could omit owner authorization ✓ Mitigated
ASH-OCLAW-873 8.8 HIGH Codex computer-use install could omit owner authorization ✓ Mitigated
ASH-OCLAW-880 8.8 HIGH Inbound voice calls could inherit owner tool authority ✓ Mitigated
ASH-OCLAW-886 8.8 HIGH Codex native tools could ignore per-chat policy ✓ Mitigated
ASH-OCLAW-887 8.8 HIGH MCP configuration changes could omit owner authorization ✓ Mitigated
ASH-OCLAW-051 8.6 HIGH Versions 2026.2.22 and 2026.2.23 contain an authorization bypass vulnerability in the s... ✓ Mitigated
ASH-OCLAW-052 8.6 HIGH Before 2026.3.12 contains an authentication bypass vulnerability in Feishu webhook mode... ✓ Mitigated
ASH-OCLAW-013 8.5 HIGH Privilege Escalation via Silent Local Auth Reconnect ✓ Mitigated
ASH-OCLAW-053 8.4 HIGH CVE-2026-25593 ✓ Mitigated
ASH-OCLAW-054 8.4 HIGH Versions prior to 2026.2.14 contain an arbitrary file read vulnerability in the exec-ap... ✓ Mitigated
ASH-OCLAW-055 8.4 HIGH Versions 2026.1.5 prior to 2026.2.12 fail to enforce mandatory authentication on the /a... ✓ Mitigated
ASH-OCLAW-056 8.4 HIGH Before 2026.3.11 contains a session sandbox escape vulnerability in the session_status ... ✓ Mitigated
ASH-OCLAW-057 8.4 HIGH Before 2026.2.17 creates session transcript JSONL files with overly broad default permi... ✓ Mitigated
ASH-OCLAW-058 8.4 HIGH Before 2026.3.12 automatically discovers and loads plugins from .OpenClaw/extensions/ w... ✓ Mitigated
ASH-OCLAW-059 8.3 HIGH Versions prior to 2026.2.14 contain server-side request forgery vulnerabilities in the ... ✓ Mitigated
ASH-OCLAW-060 8.3 HIGH Versions prior to 2026.2.14 contain a server-side request forgery vulnerability in the ... ✓ Mitigated
ASH-OCLAW-061 8.3 HIGH Before 2026.3.28 contains a server-side request forgery vulnerability in the fal provid... ✓ Mitigated
ASH-OCLAW-840 8.3 HIGH Skill tool dispatch could skip owner-only policy ✓ Mitigated
ASH-OCLAW-848 8.3 HIGH Browser node targets could bypass sandbox host control ✓ Mitigated
ASH-OCLAW-864 8.3 HIGH Agent cron tool could reach operator command jobs ✓ Mitigated
ASH-OCLAW-866 8.3 HIGH Browser proxy could miss its admin scope ✓ Mitigated
ASH-OCLAW-867 8.3 HIGH iOS Control UI did not enforce saved Gateway TLS pins ✓ Mitigated
ASH-OCLAW-001 8.2 HIGH Path Traversal in Feishu Media Download ✓ Mitigated
ASH-OCLAW-870 8.2 HIGH Browser CDP connections could discard DNS pinning ✓ Mitigated
ASH-OCLAW-010 8.1 HIGH Insufficient Access Control in Gateway /reset Endpoint ✓ Mitigated
ASH-OCLAW-062 8.1 HIGH Versions 2026.1.29-beta.1 prior to 2026.2.1 contain a path traversal vulnerability in p... ✓ Mitigated
ASH-OCLAW-063 8.1 HIGH Version 2026.1.20 prior to 2026.2.1 contains a vulnerability in the Browser Relay (exte... ✓ Mitigated
ASH-OCLAW-064 8.1 HIGH Versions prior to 2026.2.2 contain a vulnerability in the gateway WebSocket connect han... ✓ Mitigated
ASH-OCLAW-065 8.1 HIGH Versions prior to 2026.2.2 contain an authorization bypass vulnerability where clients ... ✓ Mitigated
ASH-OCLAW-066 8.1 HIGH CVE-2026-32302 ✓ Mitigated
ASH-OCLAW-067 8.1 HIGH Versions prior to 2026.2.21 contain an authentication bypass vulnerability in the Contr... ✓ Mitigated
ASH-OCLAW-068 8.1 HIGH Before 2026.3.28 contains an insufficient scope validation vulnerability in the node pa... ✓ Mitigated
ASH-OCLAW-069 8.1 HIGH Before 2026.3.28 fails to disconnect active WebSocket sessions when devices are removed... ✓ Mitigated
ASH-OCLAW-070 8.1 HIGH Before 2026.3.25 contains a privilege escalation vulnerability in the gateway plugin su... ✓ Mitigated
ASH-OCLAW-071 8.1 HIGH Before 2026.3.24 contains an incorrect authorization vulnerability in the POST /reset-p... ✓ Mitigated
ASH-OCLAW-846 8.1 HIGH WhatsApp login tool could reach non-owner turns ✓ Mitigated
ASH-OCLAW-016 8.0 HIGH Privilege Escalation via Plugin HTTP Auth Scope Minting ✓ Mitigated
ASH-OCLAW-072 8.0 HIGH Versions prior to 2026.2.26 contain a metadata spoofing vulnerability where reconnect p... ✓ Mitigated
ASH-OCLAW-073 8.0 HIGH Before 2026.3.11 contains an approval integrity vulnerability where system.run approval... ✓ Mitigated
ASH-OCLAW-845 8.0 HIGH Escaped newlines could confuse exec allowlist parsing ✓ Mitigated
ASH-OCLAW-860 8.0 HIGH Exec wrapper allowlists could trust arbitrary inner commands ✓ Mitigated
ASH-OCLAW-871 8.0 HIGH Claude permission replies could omit owner authorization ✓ Mitigated
ASH-OCLAW-074 7.8 HIGH CVE-2026-27001 ✓ Mitigated
ASH-OCLAW-075 7.8 HIGH Versions 2026.1.21 prior to 2026.2.19 contain a path hijacking vulnerability in tools.e... ✓ Mitigated
ASH-OCLAW-076 7.8 HIGH Versions prior to 2026.2.22 on macOS contain a path validation bypass vulnerability in ... ✓ Mitigated
ASH-OCLAW-077 7.8 HIGH Before 2026.3.24 contains an arbitrary code execution vulnerability in local plugin and... ✓ Mitigated
ASH-OCLAW-832 7.8 HIGH Workspace Cloud SDK arguments could execute code during Gmail setup ✓ Mitigated
ASH-OCLAW-851 7.8 HIGH OpenShell filesystem mutations could race path validation ✓ Mitigated
ASH-OCLAW-012 7.7 HIGH Path Traversal in Sandbox Enforcement (unnormalized mediaUrl) ✓ Mitigated
ASH-OCLAW-078 7.7 HIGH CVE-2026-25157 ✓ Mitigated
ASH-OCLAW-079 7.7 HIGH Versions 2.0.0-beta3 prior to 2026.2.14 contain a path traversal vulnerability in hook ... ✓ Mitigated
ASH-OCLAW-080 7.7 HIGH Versions 2026.1.29-beta.1 prior to 2026.2.14 contain a vulnerability in the sandbox bro... ✓ Mitigated
ASH-OCLAW-081 7.7 HIGH Versions prior to 2026.2.21 sandbox browser entrypoint launches x11vnc without authenti... ✓ Mitigated
ASH-OCLAW-082 7.6 HIGH CVE-2026-26322 ✓ Mitigated
ASH-OCLAW-083 7.6 HIGH CVE-2026-27487 ✓ Mitigated
ASH-OCLAW-084 7.6 HIGH Versions prior to 2026.3.2 contain a DNS pinning bypass vulnerability in strict URL fet... ✓ Mitigated
ASH-OCLAW-085 7.6 HIGH Versions prior to 2026.2.26 contain a path traversal vulnerability in workspace boundar... ✓ Mitigated
ASH-OCLAW-086 7.6 HIGH Versions prior to commit b57b680 contain an approval bypass vulnerability due to incons... ✓ Mitigated
ASH-OCLAW-826 7.6 HIGH message.action could trust spoofed requester provenance ✓ Mitigated
ASH-OCLAW-876 7.6 HIGH chat.send could expose owner-only infrastructure tools ✓ Mitigated
ASH-OCLAW-005 7.5 HIGH WebSocket Brute-Force / ClawJacked (No Rate Limiting on Localhost) ✓ Mitigated
ASH-OCLAW-007 7.5 HIGH Sandbox Escape via sessions_spawn ✓ Mitigated
ASH-OCLAW-008 7.5 HIGH Oversized Media Payload DoS ✓ Mitigated
ASH-OCLAW-014 7.5 HIGH Server-Side Request Forgery (SSRF) in Channel Extension fetch() ✓ Mitigated
ASH-OCLAW-087 7.5 HIGH CVE-2026-25474 ✓ Mitigated
ASH-OCLAW-088 7.5 HIGH CVE-2026-26316 ✓ Mitigated
ASH-OCLAW-089 7.5 HIGH CVE-2026-26319 ✓ Mitigated
ASH-OCLAW-090 7.5 HIGH CVE-2026-26321 ✓ Mitigated
ASH-OCLAW-091 7.5 HIGH CVE-2026-26324 ✓ Mitigated
ASH-OCLAW-092 7.5 HIGH Versions prior to 2026.2.14 contain a privilege escalation vulnerability in the Slack s... ✓ Mitigated
ASH-OCLAW-093 7.5 HIGH Versions prior to 2026.2.14 fail to validate TAR archive entry paths during extraction,... ✓ Mitigated
ASH-OCLAW-094 7.5 HIGH Versions prior to 2026.2.2 fail to validate webhook secrets in Telegram webhook mode (m... ✓ Mitigated
ASH-OCLAW-095 7.5 HIGH Versions prior to 2026.2.13 contain a vulnerability in the browser control API in which... ✓ Mitigated
ASH-OCLAW-096 7.5 HIGH Versions prior to 2026.2.14 contain a webhook routing vulnerability in the Google Chat ... ✓ Mitigated
ASH-OCLAW-097 7.5 HIGH Versions prior to 2026.2.13 contain a denial of service vulnerability in webhook handle... ✓ Mitigated
ASH-OCLAW-098 7.5 HIGH Versions prior to 2026.2.15 use SHA-1 to hash sandbox identifier cache keys for Docker ... ✓ Mitigated
ASH-OCLAW-099 7.5 HIGH Versions prior to 2026.2.14 contain a denial of service vulnerability in the fetchWithG... ✓ Mitigated
ASH-OCLAW-100 7.5 HIGH Versions prior to 2026.2.14 contain a local file inclusion vulnerability in BlueBubbles... ✓ Mitigated
ASH-OCLAW-101 7.5 HIGH Versions2026.2.21-2 prior to 2026.2.22 and @openclaw/voice-call versions 2026.2.21 prio... ✓ Mitigated
ASH-OCLAW-102 7.5 HIGH Versions prior to 2026.3.1 contain an unbounded memory growth vulnerability in the Zalo... ✓ Mitigated
ASH-OCLAW-103 7.5 HIGH Versions prior to 2026.3.2 contain a denial of service vulnerability in webhook handler... ✓ Mitigated
ASH-OCLAW-104 7.5 HIGH Versions prior to 2026.2.19 contain a path traversal vulnerability in the stageSandboxM... ✓ Mitigated
ASH-OCLAW-105 7.5 HIGH Versions prior to 2026.2.22 fail to sanitize shell startup environment variables HOME a... ✓ Mitigated
ASH-OCLAW-106 7.5 HIGH Through 2026.3.23 (fixed in commit 4797bbc) contains a path traversal vulnerability in ... ✓ Mitigated
ASH-OCLAW-107 7.5 HIGH Before 2026.3.13 reads and buffers Telegram webhook request bodies before validating th... ✓ Mitigated
ASH-OCLAW-108 7.5 HIGH Before 2026.3.12 embeds long-lived shared gateway credentials directly in pairing setup... ✓ Mitigated
ASH-OCLAW-109 7.5 HIGH Before 2026.3.13 contains an information disclosure vulnerability in the fetchRemoteMed... ✓ Mitigated
ASH-OCLAW-110 7.5 HIGH Before 2026.3.11 contains a sandbox boundary bypass vulnerability in fs-bridge staged w... ✓ Mitigated
ASH-OCLAW-825 7.5 HIGH Gateway upgrade-like requests could retain unauthenticated sockets ✓ Mitigated
ASH-OCLAW-835 7.5 HIGH Dreaming could widen restricted sender authority ✓ Mitigated
ASH-OCLAW-839 7.5 HIGH Reusable exec approvals could authorize changed arguments ✓ Mitigated
ASH-OCLAW-863 7.5 HIGH Matrix authorization could conflate case-distinct user IDs ✓ Mitigated
ASH-OCLAW-875 7.5 HIGH Config revision hashes could expose password verifiers ✓ Mitigated
ASH-OCLAW-111 7.4 HIGH Versions prior to 2026.3.1 contain a server-side request forgery vulnerability in web_s... ✓ Mitigated
ASH-OCLAW-112 7.4 HIGH Versions prior to 2026.2.22 contain incomplete IPv4 special-use range validation in the... ✓ Mitigated
ASH-OCLAW-113 7.4 HIGH Canvas Authentication Bypass Vulnerability ✓ Mitigated
ASH-OCLAW-015 7.3 HIGH Improper Access Control in Session Kill Endpoint ✓ Mitigated
ASH-OCLAW-114 7.3 HIGH CVE-2026-27488 ✓ Mitigated
ASH-OCLAW-115 7.3 HIGH Versions 2026.1.29 prior to 2026.2.1 contain a vulnerability in the Twitch plugin (must... ✓ Mitigated
ASH-OCLAW-116 7.3 HIGH Before 2026.3.11 contains an approval integrity vulnerability allowing attackers to exe... ✓ Mitigated
ASH-OCLAW-117 7.3 HIGH Before 2026.3.22 performs cite expansion before completing channel and DM authorization... ✓ Mitigated
ASH-OCLAW-822 7.3 HIGH Exec approvals could outlive their reviewed working directory ✓ Mitigated
ASH-OCLAW-118 7.2 HIGH CVE-2026-26325 ✓ Mitigated
ASH-OCLAW-119 7.2 HIGH Versions 2026.1.5 prior to 2026.2.14 contain a vulnerability in the Gateway in which it... ✓ Mitigated
ASH-OCLAW-120 7.2 HIGH Versions prior to 2026.2.22 in macOS node-host system.run contain an allowlist bypass v... ✓ Mitigated
ASH-OCLAW-121 7.1 HIGH CVE-2026-26317 ✓ Mitigated
ASH-OCLAW-122 7.1 HIGH Versions prior to 2026.2.12 fail to validate the sessionFile path parameter, allowing a... ✓ Mitigated
ASH-OCLAW-123 7.1 HIGH Versions prior to 2026.2.14 contain an oauth state validation bypass vulnerability in t... ✓ Mitigated
ASH-OCLAW-124 7.1 HIGH Versions prior to 2026.2.12 construct transcript file paths using unsanitized sessionId... ✓ Mitigated
ASH-OCLAW-125 7.1 HIGH Version 2026.2.19-2 prior to 2026.2.21 contains a command injection vulnerability in sy... ✓ Mitigated
ASH-OCLAW-126 7.1 HIGH Versions prior to 2026.2.22 contain an allowlist bypass vulnerability in system.run exe... ✓ Mitigated
ASH-OCLAW-127 7.1 HIGH Versions prior to 2026.2.23 contain an allowlist bypass vulnerability in system.run gua... ✓ Mitigated
ASH-OCLAW-128 7.1 HIGH Versions prior to 2026.2.19 contain a local command injection vulnerability in Windows ... ✓ Mitigated
ASH-OCLAW-129 7.1 HIGH Versions prior to 2026.2.19 contain a command injection vulnerability in the Lobster ex... ✓ Mitigated
ASH-OCLAW-130 7.1 HIGH Versions prior to 2026.2.19 contain an allowlist bypass vulnerability in the exec safeB... ✓ Mitigated
ASH-OCLAW-131 7.1 HIGH Versions prior to 2026.2.24 contain an approval gating bypass vulnerability in system.r... ✓ Mitigated
ASH-OCLAW-132 7.1 HIGH Versions prior to 2026.2.25 contain an authentication bypass vulnerability in the trust... ✓ Mitigated
ASH-OCLAW-133 7.1 HIGH Before 2026.3.11 contains an authorization bypass vulnerability allowing authenticated ... ✓ Mitigated
ASH-OCLAW-134 7.1 HIGH Before 2026.3.11 contains an approval-integrity vulnerability in node-host system.run a... ✓ Mitigated
ASH-OCLAW-135 7.1 HIGH Through 2026.2.22 contains a symlink traversal vulnerability in agents.create and agent... ✓ Mitigated
ASH-OCLAW-849 7.1 HIGH Synology file delivery could lose DNS pinning ✓ Mitigated
ASH-OCLAW-877 7.1 HIGH Signal reactions could bind to the wrong approval ✓ Mitigated
ASH-OCLAW-004 7.0 HIGH Arbitrary Shell Execution via SHELL Environment Variable ✓ Mitigated
ASH-OCLAW-136 6.9 MEDIUM Versions prior to 2026.3.1 fail to properly handle authentication bootstrap errors duri... ✓ Mitigated
ASH-OCLAW-137 6.8 MEDIUM Versions prior to 2026.2.12 with the optional Nostr plugin enabled expose unauthenticat... ✓ Mitigated
ASH-OCLAW-138 6.8 MEDIUM Versions prior to 2026.2.25 fail to enforce sender authorization checks for interactive... ✓ Mitigated
ASH-OCLAW-139 6.8 MEDIUM Versions prior to 2026.2.23 contain a path traversal vulnerability in the experimental ... ✓ Mitigated
ASH-OCLAW-861 6.8 MEDIUM Feishu tools could use disabled default accounts ✓ Mitigated
ASH-OCLAW-140 6.7 MEDIUM CVE-2026-26972 ✓ Mitigated
ASH-OCLAW-141 6.7 MEDIUM CVE-2026-27008 ✓ Mitigated
ASH-OCLAW-142 6.7 MEDIUM Versions prior to 2026.2.22 contain an allowlist bypass vulnerability in the safeBins c... ✓ Mitigated
ASH-OCLAW-143 6.7 MEDIUM 2026.3.1 contains an approval integrity vulnerability in system.run node-host execution... ✓ Mitigated
ASH-OCLAW-878 6.7 MEDIUM Windows allowlist execution could use workspace shadows ✓ Mitigated
ASH-OCLAW-144 6.6 MEDIUM Versions prior to 2026.2.22 contain an environment variable injection vulnerability in ... ✓ Mitigated
ASH-OCLAW-145 6.5 MEDIUM CVE-2026-25475 ✓ Mitigated
ASH-OCLAW-146 6.5 MEDIUM CVE-2026-26320 ✓ Mitigated
ASH-OCLAW-147 6.5 MEDIUM CVE-2026-26327 ✓ Mitigated
ASH-OCLAW-148 6.5 MEDIUM CVE-2026-26328 ✓ Mitigated
ASH-OCLAW-149 6.5 MEDIUM CVE-2026-26329 ✓ Mitigated
ASH-OCLAW-150 6.5 MEDIUM Versions prior to 2026.2.15 contain a denial of service vulnerability in the web_fetch ... ✓ Mitigated
ASH-OCLAW-151 6.5 MEDIUM Version 2026.1.14-1 prior to 2026.2.12 contain an improper network binding vulnerabilit... ✓ Mitigated
ASH-OCLAW-152 6.5 MEDIUM Versions prior to 2026.2.2 contain a server-side request forgery vulnerability in attac... ✓ Mitigated
ASH-OCLAW-153 6.5 MEDIUM Versions prior to 2026.2.14 contain an authorization bypass vulnerability where Telegra... ✓ Mitigated
ASH-OCLAW-154 6.5 MEDIUM Versions 2026.1.30 and earlier, contain an information disclosure vulnerability, patche... ✓ Mitigated
ASH-OCLAW-155 6.5 MEDIUM Versions prior to 2026.2.14 contain a webhook signature-verification bypass in the voic... ✓ Mitigated
ASH-OCLAW-156 6.5 MEDIUM Versions prior to 2026.2.21 contain an approval-integrity mismatch vulnerability in sys... ✓ Mitigated
ASH-OCLAW-157 6.5 MEDIUM Versions prior to 2026.2.22 with the optional BlueBubbles plugin contain an access cont... ✓ Mitigated
ASH-OCLAW-158 6.5 MEDIUM Versions prior to 2026.2.19 construct RegExp objects directly from unescaped Feishu men... ✓ Mitigated
ASH-OCLAW-159 6.5 MEDIUM Versions prior to 2026.2.24 contain a local media root bypass vulnerability in sendAtta... ✓ Mitigated
ASH-OCLAW-160 6.5 MEDIUM Versions prior to 2026.2.25 lack durable replay state for Nextcloud Talk webhook events... ✓ Mitigated
ASH-OCLAW-161 6.5 MEDIUM Versions prior to 2026.3.2 contain an authentication bypass vulnerability in the /api/c... ✓ Mitigated
ASH-OCLAW-162 6.5 MEDIUM Versions prior to 2026.2.21 contain an improper URL scheme validation vulnerability in ... ✓ Mitigated
ASH-OCLAW-163 6.5 MEDIUM Versions prior to 2026.2.22 contain an authorization bypass vulnerability in the Feishu... ✓ Mitigated
ASH-OCLAW-164 6.5 MEDIUM Versions prior to 2026.2.21 contain a stdin-only policy bypass vulnerability in the gre... ✓ Mitigated
ASH-OCLAW-165 6.5 MEDIUM Versions prior to 2026.2.24 contain an improper path validation vulnerability in sandbo... ✓ Mitigated
ASH-OCLAW-166 6.5 MEDIUM Versions prior to 2026.2.26 contain an authorization bypass vulnerability where DM pair... ✓ Mitigated
ASH-OCLAW-167 6.5 MEDIUM Versions prior to 2026.2.24 contain a path traversal vulnerability where @-prefixed abs... ✓ Mitigated
ASH-OCLAW-168 6.5 MEDIUM Gateway plugin versions prior to 2026.2.26 contain a path traversal vulnerability that ... ✓ Mitigated
ASH-OCLAW-169 6.5 MEDIUM Versions prior to 2026.2.25 contain a time-of-check-time-of-use vulnerability in approv... ✓ Mitigated
ASH-OCLAW-170 6.5 MEDIUM Versions prior to 2026.2.23 contain a vulnerability in Twilio webhook event deduplicati... ✓ Mitigated
ASH-OCLAW-171 6.5 MEDIUM Versions prior to 2026.2.25 contain a symlink traversal vulnerability in browser trace ... ✓ Mitigated
ASH-OCLAW-172 6.5 MEDIUM Before 2026.3.11 contains an authorization bypass vulnerability allowing channel comman... ✓ Mitigated
ASH-OCLAW-173 6.5 MEDIUM Before 2026.3.12 applies rate limiting only after successful webhook authentication, al... ✓ Mitigated
ASH-OCLAW-174 6.5 MEDIUM Before 2026.3.28 downloads and stores inbound media from Zalo channels before validatin... ✓ Mitigated
ASH-OCLAW-175 6.5 MEDIUM Before 2026.3.28 contains a missing rate limiting vulnerability in the Nextcloud Talk w... ✓ Mitigated
ASH-OCLAW-176 6.5 MEDIUM Before 2026.3.24 contains a sandbox bypass vulnerability in the message tool that allow... ✓ Mitigated
ASH-OCLAW-177 6.5 MEDIUM Before 2026.3.31 (patched in 2026.4.8) contains a request body replay vulnerability in ... ✓ Mitigated
ASH-OCLAW-178 6.5 MEDIUM Before 2026.3.23 contains a replay identity vulnerability in Plivo V2 signature verific... ✓ Mitigated
ASH-OCLAW-179 6.5 MEDIUM Before 2026.3.22 performs cryptographic and dispatch operations on inbound Nostr direct... ✓ Mitigated
ASH-OCLAW-180 6.5 MEDIUM Before 2026.3.22 fails to enforce operator.admin scope on mutating internal ACP chat co... ✓ Mitigated
ASH-OCLAW-181 6.5 MEDIUM Versions 2026.3.11 through 2026.3.24 contain a session isolation bypass vulnerability w... ✓ Mitigated
ASH-OCLAW-182 6.5 MEDIUM Before 2026.3.22 contains an information disclosure vulnerability that allows attackers... ✓ Mitigated
ASH-OCLAW-183 6.5 MEDIUM Before 2026.3.24 contains a privilege escalation vulnerability where the /allowlist com... ✓ Mitigated
ASH-OCLAW-184 6.5 MEDIUM Before 2026.3.22 contains a settings reconciliation vulnerability that allows attackers... ✓ Mitigated
ASH-OCLAW-185 6.5 MEDIUM Before 2026.3.22 contains an authorization bypass vulnerability in interactive callback... ✓ Mitigated
ASH-OCLAW-186 6.5 MEDIUM Before 2026.3.22 contains an authentication bypass vulnerability in the X-Forwarded-For... ✓ Mitigated
ASH-OCLAW-187 6.5 MEDIUM Before 2026.3.25 contains an authorization bypass vulnerability in the HTTP /sessions/:... ✓ Mitigated
ASH-OCLAW-188 6.5 MEDIUM Before 2026.3.2 contains a filesystem boundary bypass vulnerability in the image tool t... ✓ Mitigated
ASH-OCLAW-189 6.5 MEDIUM Canvas Path Traversal Information Disclosure Vulnerability ✓ Mitigated
ASH-OCLAW-824 6.5 MEDIUM Diagnostics export could omit owner authorization ✓ Mitigated
ASH-OCLAW-843 6.5 MEDIUM Trajectory export could omit owner authorization ✓ Mitigated
ASH-OCLAW-852 6.5 MEDIUM LINE group allowlist could inherit DM access ✓ Mitigated
ASH-OCLAW-853 6.5 MEDIUM Structured attachments could hide unvalidated host paths ✓ Mitigated
ASH-OCLAW-858 6.5 MEDIUM Channel read actions could skip target allowlists ✓ Mitigated
ASH-OCLAW-882 6.5 MEDIUM Slack file downloads could miss conversation authorization ✓ Mitigated
ASH-OCLAW-884 6.5 MEDIUM Outbound attachments could ignore requester read denials ✓ Mitigated
ASH-OCLAW-003 6.4 MEDIUM Allow-Always Wrapper Persistence Bypass ✓ Mitigated
ASH-OCLAW-190 6.4 MEDIUM Versions prior to 2026.2.24 contain a command injection vulnerability in the system.run... ✓ Mitigated
ASH-OCLAW-834 6.4 MEDIUM Discord realtime transcripts could inherit another speaker's owner status ✓ Mitigated
ASH-OCLAW-841 6.4 MEDIUM File-transfer approvals could widen durable authority ✓ Mitigated
ASH-OCLAW-011 6.3 MEDIUM Environment Variable Override Bypass ✓ Mitigated
ASH-OCLAW-191 6.3 MEDIUM A vulnerability was determined in OpenClaw 2026.2.19-2 ✓ Mitigated
ASH-OCLAW-192 6.3 MEDIUM Versions 2026.2.26 prior to 2026.3.1 on Windows contain a current working directory inj... ✓ Mitigated
ASH-OCLAW-193 6.3 MEDIUM Versions prior to 2026.2.22 contain an allowlist bypass vulnerability in the safe-bin c... ✓ Mitigated
ASH-OCLAW-194 6.3 MEDIUM Before 2026.3.8 contains an approval bypass vulnerability in system.run where mutable s... ✓ Mitigated
ASH-OCLAW-195 6.3 MEDIUM Before 2026.3.11 contains a sandbox boundary bypass vulnerability in the fs-bridge writ... ✓ Mitigated
ASH-OCLAW-818 6.3 MEDIUM Memory dreaming changes could omit owner checks ✓ Mitigated
ASH-OCLAW-830 6.3 MEDIUM iOS deep-link logs could expose unattended credentials ✓ Mitigated
ASH-OCLAW-854 6.3 MEDIUM Video asset downloads could reach private destinations ✓ Mitigated
ASH-OCLAW-869 6.3 MEDIUM WhatsApp session resets could bypass command authorization ✓ Mitigated
ASH-OCLAW-890 6.3 MEDIUM Active Memory global toggles could omit owner checks ✓ Mitigated
ASH-OCLAW-196 6.2 MEDIUM Before 2026.3.8 contains a path traversal vulnerability in the skills download installe... ✓ Mitigated
ASH-OCLAW-197 6.1 MEDIUM Versions prior to 2026.2.14 contain a path traversal vulnerability in sandbox skill mir... ✓ Mitigated
ASH-OCLAW-198 6.1 MEDIUM Versions 2026.1.16-2 prior to 2026.2.14 contain a path traversal vulnerability in archi... ✓ Mitigated
ASH-OCLAW-199 6.1 MEDIUM Version 2026.2.22 prior to 2026.2.23 contain an arbitrary code execution vulnerability ... ✓ Mitigated
ASH-OCLAW-200 6.1 MEDIUM Versions prior to 2026.2.24 contain a sandbox bind validation vulnerability allowing at... ✓ Mitigated
ASH-OCLAW-201 6.1 MEDIUM Versions prior to 2026.2.26 contain an approval bypass vulnerability in system.run exec... ✓ Mitigated
ASH-OCLAW-202 6.1 MEDIUM Versions prior to 2026.3.2 contain a vulnerability in the stageSandboxMedia function in... ✓ Mitigated
ASH-OCLAW-203 6.1 MEDIUM Versions prior to 2026.3.7 contain a sandbox escape vulnerability in the /acp spawn com... ✓ Mitigated
ASH-OCLAW-204 6.1 MEDIUM Before 2026.3.11 contains an authorization bypass vulnerability allowing write-scoped c... ✓ Mitigated
ASH-OCLAW-205 6.1 MEDIUM Before 2026.3.24 contains an incomplete fix for CVE-2026-27486 where the !stop chat com... ✓ Mitigated
ASH-OCLAW-206 6.0 MEDIUM Versions prior to 2026.3.1 fail to pin executable identity for non-path-like argv[0] to... ✓ Mitigated
ASH-OCLAW-207 6.0 MEDIUM Versions prior to 2026.2.22 fail to consistently validate redirect chains against confi... ✓ Mitigated
ASH-OCLAW-002 5.9 MEDIUM Allowlist Bypass via Shell Line-Continuation ✓ Mitigated
ASH-OCLAW-208 5.9 MEDIUM Versions prior to 2026.2.12 use non-constant-time string comparison for hook token vali... ✓ Mitigated
ASH-OCLAW-209 5.9 MEDIUM OpenClaw's voice-call plugin versions before 2026.2.3 contain an improper authenticatio... ✓ Mitigated
ASH-OCLAW-210 5.9 MEDIUM Versions prior to 2026.2.12 contain a vulnerability in the BlueBubbles (optional plugin... ✓ Mitigated
ASH-OCLAW-211 5.9 MEDIUM Versions prior to 2026.3.2 fail to pass the senderIsOwner flag when processing Discord ... ✓ Mitigated
ASH-OCLAW-212 5.9 MEDIUM Versions prior to 2026.2.22 contain an authorization bypass vulnerability in the toolsB... ✓ Mitigated
ASH-OCLAW-213 5.9 MEDIUM Versions prior to 2026.2.21 incorrectly apply tokenless Tailscale header authentication... ✓ Mitigated
ASH-OCLAW-214 5.9 MEDIUM Before 2026.3.22 contains an improper authentication verification vulnerability in Goog... ✓ Mitigated
ASH-OCLAW-215 5.9 MEDIUM Before 2026.3.22 contains a webhook reply delivery vulnerability that allows attackers ... ✓ Mitigated
ASH-OCLAW-838 5.9 MEDIUM Trusted-host DNS checks could admit unspecified addresses ✓ Mitigated
ASH-OCLAW-216 5.8 MEDIUM CVE-2026-27009 ✓ Mitigated
ASH-OCLAW-217 5.7 MEDIUM Versions prior to 2026.2.24 contain a policy bypass vulnerability in the safeBins allow... ✓ Mitigated
ASH-OCLAW-218 5.7 MEDIUM Before 2026.3.22 contains an identity spoofing vulnerability in ACP permission resoluti... ✓ Mitigated
ASH-OCLAW-219 5.6 MEDIUM 1.26 ✓ Mitigated
ASH-OCLAW-220 5.5 MEDIUM CVE-2026-27003 ✓ Mitigated
ASH-OCLAW-221 5.5 MEDIUM CVE-2026-27004 ✓ Mitigated
ASH-OCLAW-222 5.5 MEDIUM Versions prior to 2026.2.14 contain a denial of service vulnerability in the extractArc... ✓ Mitigated
ASH-OCLAW-223 5.5 MEDIUM Versions prior to 2026.2.14 decode base64-backed media inputs into buffers before enfor... ✓ Mitigated
ASH-OCLAW-224 5.5 MEDIUM Versions prior to 2026.2.22 contain a symlink traversal vulnerability in avatar handlin... ✓ Mitigated
ASH-OCLAW-225 5.5 MEDIUM Versions prior to 2026.3.2 contain an archive extraction vulnerability in the tar.bz2 i... ✓ Mitigated
ASH-OCLAW-819 5.5 MEDIUM ACP file URLs could skip out-of-cwd read approval ✓ Mitigated
ASH-OCLAW-859 5.5 MEDIUM iOS Share Extension stored Gateway credentials outside Keychain ✓ Mitigated
ASH-OCLAW-872 5.5 MEDIUM Workspace endpoint override could expose Azure Speech credentials ✓ Mitigated
ASH-OCLAW-226 5.4 MEDIUM Versions prior to 2026.2.22 contain an authentication bypass vulnerability that allows ... ✓ Mitigated
ASH-OCLAW-227 5.4 MEDIUM Versions prior to 2026.2.26 fail to enforce sender authorization in member and message ... ✓ Mitigated
ASH-OCLAW-228 5.4 MEDIUM Versions prior to 2026.2.23 contain an authorization bypass vulnerability in the ACP cl... ✓ Mitigated
ASH-OCLAW-229 5.4 MEDIUM Before 2026.3.11 contains an authorization bypass vulnerability in Discord guild reacti... ✓ Mitigated
ASH-OCLAW-230 5.4 MEDIUM Versions prior to commit 8aceaf5 contain a preflight validation bypass vulnerability in... ✓ Mitigated
ASH-OCLAW-231 5.4 MEDIUM Before 2026.3.24 contains missing authorization vulnerabilities in the /send and /allow... ✓ Mitigated
ASH-OCLAW-821 5.4 MEDIUM Google Meet attendance CSV could interpret names as formulas ✓ Mitigated
ASH-OCLAW-823 5.4 MEDIUM Browser wait predicates could reach blocked destinations ✓ Mitigated
ASH-OCLAW-842 5.4 MEDIUM Microsoft Teams access-group failures could admit unlisted senders ✓ Mitigated
ASH-OCLAW-865 5.4 MEDIUM sessions.create could miss admin scope for keyed reconfiguration ✓ Mitigated
ASH-OCLAW-874 5.4 MEDIUM Group activation changes could omit owner authorization ✓ Mitigated
ASH-OCLAW-879 5.4 MEDIUM QQBot voice filenames could escape the staging directory ✓ Mitigated
ASH-OCLAW-883 5.4 MEDIUM OpenAI-compatible transport could send provider credentials to the wrong endpoint ✓ Mitigated
ASH-OCLAW-891 5.4 MEDIUM Usage CSV export could interpret session labels as formulas ✓ Mitigated
ASH-OCLAW-232 5.3 MEDIUM CVE-2026-27486 ✓ Mitigated
ASH-OCLAW-233 5.3 MEDIUM Version 2026.1.14-1 prior to 2026.2.2, with the Matrix plugin installed and enabled, co... ✓ Mitigated
ASH-OCLAW-234 5.3 MEDIUM Versions prior to 2026.3.2 contain a path-confinement bypass vulnerability in browser o... ✓ Mitigated
ASH-OCLAW-235 5.3 MEDIUM Versions prior to 2026.3.2 contain a race condition vulnerability in ZIP extraction tha... ✓ Mitigated
ASH-OCLAW-236 5.3 MEDIUM Versions 2026.1.21 prior to 2026.2.19 contain a command injection vulnerability in the ... ✓ Mitigated
ASH-OCLAW-237 5.3 MEDIUM Versions prior to 2026.2.23 contain a sandbox bypass vulnerability in the sandboxed ima... ✓ Mitigated
ASH-OCLAW-238 5.3 MEDIUM Versions prior to 2026.2.25 fail to enforce dmPolicy and allowFrom authorization checks... ✓ Mitigated
ASH-OCLAW-239 5.3 MEDIUM Versions prior to 2026.2.21 improperly parse the left-most X-Forwarded-For header value... ✓ Mitigated
ASH-OCLAW-240 5.3 MEDIUM Versions prior to 2026.2.21 contain an improper sandbox configuration vulnerability tha... ✓ Mitigated
ASH-OCLAW-241 5.3 MEDIUM Versions prior to 2026.3.7 contain a shell approval gating bypass vulnerability in syst... ✓ Mitigated
ASH-OCLAW-242 5.3 MEDIUM Before 2026.3.22 contains a path traversal vulnerability in Windows media loaders that ... ✓ Mitigated
ASH-OCLAW-243 5.3 MEDIUM Before 2026.4.2 reuses the PKCE verifier as the OAuth state parameter in the Gemini OAu... ✓ Mitigated
ASH-OCLAW-244 5.3 MEDIUM Before 2026.3.22 contains an unauthenticated resource exhaustion vulnerability in voice... ✓ Mitigated
ASH-OCLAW-245 5.3 MEDIUM Before 2026.3.22 contains an unbounded memory allocation vulnerability in remote media ... ✓ Mitigated
ASH-OCLAW-246 5.3 MEDIUM Before 2026.3.25 parses JSON request bodies before validating webhook signatures, allow... ✓ Mitigated
ASH-OCLAW-247 5.3 MEDIUM Before 2026.3.25 contains an access control vulnerability where verification notices by... ✓ Mitigated
ASH-OCLAW-248 5.3 MEDIUM Before 2026.3.25 contains an authorization bypass vulnerability in Microsoft Teams feed... ✓ Mitigated
ASH-OCLAW-249 5.3 MEDIUM Before 2026.3.25 contains an authorization bypass vulnerability in Telegram callback qu... ✓ Mitigated
ASH-OCLAW-250 5.3 MEDIUM Before 2026.3.25 contains an authentication bypass vulnerability in raw card send surfa... ✓ Mitigated
ASH-OCLAW-251 5.3 MEDIUM Before 2026.3.24 contains an incomplete fix for CVE-2026-32011 where the Feishu webhook... ✓ Mitigated
ASH-OCLAW-252 5.3 MEDIUM Client PKCE Verifier Information Disclosure Vulnerability ✓ Mitigated
ASH-OCLAW-828 5.3 MEDIUM Session filename generation could regain denied tools ✓ Mitigated
ASH-OCLAW-833 5.3 MEDIUM Unicode fallback could escape workspaceOnly roots ✓ Mitigated
ASH-OCLAW-836 5.3 MEDIUM Embedding fallback could forward a provider key across vendors ✓ Mitigated
ASH-OCLAW-855 5.3 MEDIUM Synology pre-auth limits could lock out valid webhooks ✓ Mitigated
ASH-OCLAW-868 5.3 MEDIUM Browser relay authentication could exhaust shared pending capacity ✓ Mitigated
ASH-OCLAW-888 5.3 MEDIUM Discord asset uploads could skip sender media policy ✓ Mitigated
ASH-OCLAW-889 5.3 MEDIUM Twilio pre-auth limits could lock out valid webhooks ✓ Mitigated
ASH-OCLAW-253 5.1 MEDIUM Before 2026.3.23 contains an authentication bypass vulnerability in the Canvas gateway ... ✓ Mitigated
ASH-OCLAW-254 4.8 MEDIUM Versions prior to 2026.2.13 use non-constant-time string comparison for hook token vali... ✓ Mitigated
ASH-OCLAW-255 4.8 MEDIUM Versions prior to 2026.2.22 contain an allowlist parsing mismatch vulnerability in the ... ✓ Mitigated
ASH-OCLAW-256 4.8 MEDIUM Versions prior to 2026.2.26 server-http contains an authentication bypass vulnerability... ✓ Mitigated
ASH-OCLAW-257 4.8 MEDIUM Versions prior to 2026.2.25 contain an approval-integrity bypass vulnerability in syste... ✓ Mitigated
ASH-OCLAW-258 4.8 MEDIUM Versions prior to 2026.2.21 BlueBubbles webhook handler contains a passwordless fallbac... ✓ Mitigated
ASH-OCLAW-259 4.8 MEDIUM Before 2026.3.25 contains a missing rate limiting vulnerability in webhook authenticati... ✓ Mitigated
ASH-OCLAW-260 4.8 MEDIUM Before 2026.3.25 contains a missing rate limiting vulnerability in Telegram webhook aut... ✓ Mitigated
ASH-OCLAW-261 4.8 MEDIUM Before 2026.3.22 contains a webhook path route replacement vulnerability in the Synolog... ✓ Mitigated
ASH-OCLAW-262 4.8 MEDIUM Before 2026.3.25 contains a pre-authentication rate-limit bypass vulnerability in webho... ✓ Mitigated
ASH-OCLAW-263 4.6 MEDIUM Versions prior to 2026.2.23 contain an html injection vulnerability in the HTML session... ✓ Mitigated
ASH-OCLAW-264 4.6 MEDIUM Before 2026.3.22 contains a service discovery vulnerability where TXT metadata from Bon... ✓ Mitigated
ASH-OCLAW-265 4.4 MEDIUM CVE-2026-27485 ✓ Mitigated
ASH-OCLAW-266 4.4 MEDIUM Versions prior to 2026.2.17 contain a path traversal vulnerability in the $include dire... ✓ Mitigated
ASH-OCLAW-267 4.4 MEDIUM Versions prior to 2026.2.19 tools.exec.safeBins contains an input validation bypass vul... ✓ Mitigated
ASH-OCLAW-268 4.3 MEDIUM CVE-2026-26326 ✓ Mitigated
ASH-OCLAW-269 4.3 MEDIUM CVE-2026-27484 ✓ Mitigated
ASH-OCLAW-270 4.3 MEDIUM Versions prior to 2026.2.21 accept prototype-reserved keys in runtime /debug set overri... ✓ Mitigated
ASH-OCLAW-271 4.3 MEDIUM Versions prior to 2026.2.25 fail to consistently apply sender-policy checks to reaction... ✓ Mitigated
ASH-OCLAW-272 4.3 MEDIUM Before 2026.3.8 contains a sender allowlist bypass vulnerability in its Microsoft Teams... ✓ Mitigated
ASH-OCLAW-273 4.3 MEDIUM Before 2026.3.28 contains a sender policy bypass vulnerability in the Google Chat and Z... ✓ Mitigated
ASH-OCLAW-274 4.3 MEDIUM Before 2026.3.25 contains an authorization bypass vulnerability where group reaction ev... ✓ Mitigated
ASH-OCLAW-275 4.3 MEDIUM Before 2026.3.24 contains an authorization bypass vulnerability in the HTTP /v1/models ... ✓ Mitigated
ASH-OCLAW-276 4.3 MEDIUM Versions 2026.2.13 through 2026.3.24 contain an ANSI escape sequence injection vulnerab... ✓ Mitigated
ASH-OCLAW-277 4.3 MEDIUM Before 2026.3.22 fails to enforce controlScope restrictions on the send action, allowin... ✓ Mitigated
ASH-OCLAW-820 4.3 MEDIUM Discord guild reads could skip target allowlists ✓ Mitigated
ASH-OCLAW-829 4.3 MEDIUM Talk Voice changes could omit admin authorization ✓ Mitigated
ASH-OCLAW-862 4.3 MEDIUM Feishu unpin could miss cross-context policy ✓ Mitigated
ASH-OCLAW-881 4.3 MEDIUM Prometheus diagnostics could omit operator.read enforcement ✓ Mitigated
ASH-OCLAW-278 4.2 MEDIUM Before 2026.3.25 contains an authorization bypass vulnerability in Google Chat group po... ✓ Mitigated
ASH-OCLAW-279 4.2 MEDIUM Before 2026.3.22 contains a policy confusion vulnerability in room authorization that m... ✓ Mitigated
ASH-OCLAW-857 4.2 MEDIUM Revoked Canvas capability could remain active during disconnect ✓ Mitigated
ASH-OCLAW-280 4.0 MEDIUM CVE-2026-27576 ✓ Mitigated
ASH-OCLAW-281 3.7 LOW (formerly Clawdbot) is a personal AI assistant users run on their own devices ✓ Mitigated
ASH-OCLAW-282 3.7 LOW Versions prior to 2026.2.26 contain an authorization bypass vulnerability where Signal ... ✓ Mitigated
ASH-OCLAW-283 3.7 LOW Versions prior to 2026.2.25 contain an access control vulnerability in signal reaction ... ✓ Mitigated
ASH-OCLAW-284 3.7 LOW Versions prior to 2026.2.26 contains an authorization bypass vulnerability in the pairi... ✓ Mitigated
ASH-OCLAW-285 3.7 LOW Versions prior to 2026.2.22 reuse gateway.auth.token as a fallback hash secret for owne... ✓ Mitigated
ASH-OCLAW-286 3.7 LOW Before 2026.3.22 contains a policy bypass vulnerability where queued node actions are n... ✓ Mitigated
ASH-OCLAW-287 3.6 LOW Versions prior to 2026.2.19 contain a race condition vulnerability in concurrent update... ✓ Mitigated
ASH-OCLAW-288 3.3 LOW CVE-2026-27007 ✓ Mitigated
ASH-OCLAW-289 3.3 LOW 2.17 ✓ Mitigated
ASH-OCLAW-290 3.3 LOW Versions prior to 2026.2.22 contain a path traversal vulnerability in the static file h... ✓ Mitigated
ASH-OCLAW-831 3.3 LOW MCP loopback could omit sandbox tool deny policy ✓ Mitigated
ASH-OCLAW-291 3.1 LOW Versions prior to 2026.2.26 contain an authorization bypass vulnerability where DM pair... ✓ Mitigated
ASH-OCLAW-837 3.1 LOW Truncated tar listings could bypass extraction limits ✓ Mitigated
ASH-OCLAW-856 3.1 LOW Webhook TaskFlow cancellation could target unrelated sessions ✓ Mitigated
ASH-OCLAW-885 3.1 LOW Active Memory recall could ignore requester policy ✓ Mitigated
ASH-OCLAW-292 2.6 LOW Versions prior to 2026.2.26 contain an approval context-binding weakness in system.run ... ✓ Mitigated
ASH-OCLAW-850 2.6 LOW Memory tools could retain revoked access ✓ Mitigated
ASH-OCLAW-293 2.5 LOW Before 2026.3.11 contains a credential fallback vulnerability where unavailable local g... ✓ Mitigated
ASH-OCLAW-294 — HIGH 1-Click RCE via Authentication Token Exfiltration From gatewayUrl ✓ Mitigated
ASH-OCLAW-295 — HIGH OC-02: Gateway /tools/invoke tool escalation + ACP permission auto-approval ✓ Mitigated
ASH-OCLAW-296 — CRITICAL Remote Code Execution via Node Invoke Approval Bypass in Gateway ✓ Mitigated
ASH-OCLAW-297 — HIGH Hook Session Key Override Enables Targeted Cross-Session Routing ✓ Mitigated
ASH-OCLAW-298 — HIGH Arbitrary transcript path file write via gateway sessionFile ✓ Mitigated
ASH-OCLAW-299 — HIGH SSRF in Image Tool Remote Fetch ✓ Mitigated
ASH-OCLAW-300 — MEDIUM Webhook auth bypass when gateway is behind a reverse proxy (loopback remoteAddress trust) ✓ Mitigated
ASH-OCLAW-301 — MEDIUM SSRF via attachment/media URL hydration ✓ Mitigated
ASH-OCLAW-302 — HIGH Unauthenticated Nostr profile HTTP endpoints allow remote profile/config tampering ✓ Mitigated
ASH-OCLAW-303 — LOW OpenClaw log poisoning (indirect prompt injection) via WebSocket headers ✓ Mitigated
ASH-OCLAW-304 — MEDIUM Optional voice-call plugin: webhook verification may be bypassed behind certain proxy conf ✓ Mitigated
ASH-OCLAW-305 — HIGH Potential access-group authorization bypass if channel type lookup fails ✓ Mitigated
ASH-OCLAW-306 — HIGH Nextcloud Talk allowlist bypass via actor.name display name spoofing ✓ Mitigated
ASH-OCLAW-307 — HIGH Authorization bypass: `operator.write` can resolve exec approvals via `chat.send` -> `/app ✓ Mitigated
ASH-OCLAW-308 — CRITICAL Path Traversal in Plugin Installation ✓ Mitigated
ASH-OCLAW-309 — HIGH Gateway connect could skip device identity checks when auth.token was present but not yet ✓ Mitigated
ASH-OCLAW-310 — HIGH Windows cmd.exe parsing may bypass exec allowlist/approval gating ✓ Mitigated
ASH-OCLAW-311 — HIGH Browser Relay /cdp websocket missing auth could allow cross-tab cookie access ✓ Mitigated
ASH-OCLAW-312 — HIGH Exec allowlist bypass via command substitution/backticks inside double quotes ✓ Mitigated
ASH-OCLAW-313 — MEDIUM Slack: dmPolicy=open allowed any DM sender to run privileged slash commands ✓ Mitigated
ASH-OCLAW-314 — HIGH Denial of service via unbounded webhook request body buffering ✓ Mitigated
ASH-OCLAW-315 — HIGH Google Chat shared-path webhook target ambiguity allowed cross-account policy-context misr ✓ Mitigated
ASH-OCLAW-316 — MEDIUM SSRF in optional Tlon (Urbit) extension authentication ✓ Mitigated
ASH-OCLAW-317 — MEDIUM Non-constant-time token comparison in hooks authentication ✓ Mitigated
ASH-OCLAW-318 — MEDIUM Config writes could persist resolved ${VAR} secrets to disk ✓ Mitigated
ASH-OCLAW-319 — MEDIUM Inter-session prompts could be treated as direct user instructions ✓ Mitigated
ASH-OCLAW-320 — HIGH Path traversal (Zip Slip) in archive extraction during explicit installation commands ✓ Mitigated
ASH-OCLAW-321 — LOW Google Chat spoofing access with allowlist authorized mutable email principal despite send ✓ Mitigated
ASH-OCLAW-322 — HIGH missing authentication for local browser-control endpoints ✓ Mitigated
ASH-OCLAW-323 — HIGH Potential code execution via unsafe hook module path handling in Gateway ✓ Mitigated
ASH-OCLAW-324 — HIGH Zip Slip path traversal in tar archive extraction ✓ Mitigated
ASH-OCLAW-325 — HIGH LFI in BlueBubbles media path handling ✓ Mitigated
ASH-OCLAW-326 — HIGH Two SSRF via sendMediaFeishu and markdown image fetching in Feishu extension ✓ Mitigated
ASH-OCLAW-327 — MEDIUM Hook transform module path allows traversal and arbitrary JavaScript module loading ✓ Mitigated
ASH-OCLAW-328 — LOW Option injection in pre-commit hook can stage ignored files ✓ Mitigated
ASH-OCLAW-329 — HIGH Path traversal in apply_patch could write/delete files outside the workspace ✓ Mitigated
ASH-OCLAW-330 — MEDIUM OpenClaw safeBins file-existence oracle information disclosure ✓ Mitigated
ASH-OCLAW-331 — MEDIUM Windows Scheduled Task script generation allowed local command injection via unsafe cmd ar ✓ Mitigated
ASH-OCLAW-332 — MEDIUM Command Injection via unescaped environment assignments in Windows Scheduled Task script g ✓ Mitigated
ASH-OCLAW-333 — LOW Windows Lobster shell fallback command injection in constrained fallback path ✓ Mitigated
ASH-OCLAW-334 — MEDIUM Auth inconsistency on local Browser Extension Relay /extension endpoint ✓ Mitigated
ASH-OCLAW-335 — MEDIUM SSRF guard bypass via IPv6 transition over ISATAP ✓ Mitigated
ASH-OCLAW-336 — MEDIUM iMessage remote attachment SCP hardening (strict host-key checks and remoteHost validation ✓ Mitigated
ASH-OCLAW-337 — MEDIUM Path traversal in Feishu media temp-file naming allows writes outside os.tmpdir() ✓ Mitigated
ASH-OCLAW-338 — MEDIUM Command injection via Windows shell fallback in Lobster tool execution ✓ Mitigated
ASH-OCLAW-339 — LOW Plugin runtime command execution is part of trusted plugin boundary ✓ Mitigated
ASH-OCLAW-340 — LOW Owner-only gateway tool access checks were incomplete in specific authenticated DM flows ✓ Mitigated
ASH-OCLAW-341 — MEDIUM OpenClaw exec allowlist safeBins short-option bypass could permit arbitrary file write ✓ Mitigated
ASH-OCLAW-342 — MEDIUM `tools.exec.safeBins` PATH-hijack allowed trojan binaries to bypass allowlist checks ✓ Mitigated
ASH-OCLAW-343 — HIGH ZDI-CAN-29311: OpenClaw Canvas Authentication Bypass Vulnerability ✓ Mitigated
ASH-OCLAW-344 — MEDIUM ZDI-CAN-29312: OpenClaw Canvas Path Traversal Information Disclosure Vulnerability ✓ Mitigated
ASH-OCLAW-345 — MEDIUM Local File Exfiltration via MCP Tool Result MEDIA: Directive Injection ✓ Mitigated
ASH-OCLAW-346 — MEDIUM Skill env override host env injection ✓ Mitigated
ASH-OCLAW-347 — HIGH OC-06: Arbitrary file read via $include directive ✓ Mitigated
ASH-OCLAW-348 — HIGH Multiple E2E/test Dockerfiles run all processes as root ✓ Mitigated
ASH-OCLAW-349 — MEDIUM Browser navigation guard allowed non-network URL schemes, enabling authenticated browser-t ✓ Mitigated
ASH-OCLAW-350 — MEDIUM Windows system.run approval mismatch on cmd.exe /c trailing arguments ✓ Mitigated
ASH-OCLAW-351 — MEDIUM BlueBubbles beta plugin webhook auth hardening (remove passwordless fallback) ✓ Mitigated
ASH-OCLAW-352 — MEDIUM BASH_ENV / ENV startup-file injection into spawned shell commands ✓ Mitigated
ASH-OCLAW-353 — MEDIUM Config env vars allowed startup env injection into service runtime ✓ Mitigated
ASH-OCLAW-354 — MEDIUM Chrome --no-sandbox disabled OS-level browser sandbox in sandbox browser container ✓ Mitigated
ASH-OCLAW-355 — MEDIUM Canvas route hardening for mixed-trust deployments ✓ Mitigated
ASH-OCLAW-356 — HIGH Shell command injection in exec allowlist mode via unquoted heredoc expansion ✓ Mitigated
ASH-OCLAW-357 — MEDIUM OC-91: WhatsApp reaction action bypasses JID authorization for allowFrom-restricted target ✓ Mitigated
ASH-OCLAW-358 — MEDIUM TTS model directives allowed provider switching by default ✓ Mitigated
ASH-OCLAW-359 — MEDIUM ACP resource_link metadata prompt interpolation allowed prompt-injection content via title ✓ Mitigated
ASH-OCLAW-360 — LOW OC-65: overflow compaction retry counter reset after truncation amplifies bounded retry/co ✓ Mitigated
ASH-OCLAW-361 — MEDIUM tools.exec.safeBins trusted PATH directories allowed binary shadowing in allowlist mode ✓ Mitigated
ASH-OCLAW-362 — MEDIUM Non-default safeBins sort configuration can bypass intended allowlist approval constraints ✓ Mitigated
ASH-OCLAW-363 — MEDIUM Avatar symlink traversal can expose out-of-workspace local files ✓ Mitigated
ASH-OCLAW-364 — MEDIUM allow-always wrapper persistence could bypass future approvals and enable command executio ✓ Mitigated
ASH-OCLAW-365 — MEDIUM system.run allowlist bypass via shell line-continuation command substitution ✓ Mitigated
ASH-OCLAW-366 — MEDIUM Elevated allowFrom matching tightened for sender-scoped authorization ✓ Mitigated
ASH-OCLAW-367 — MEDIUM Feishu allowFrom authorization bypass via display-name collision ✓ Mitigated
ASH-OCLAW-368 — MEDIUM Shell-env fallback trusted startup env and could execute attacker-influenced login-shell p ✓ Mitigated
ASH-OCLAW-369 — LOW tools.exec.safeBins generic fallback allowed interpreter-style inline payload execution in ✓ Mitigated
ASH-OCLAW-370 — MEDIUM system.run shell-wrapper env injection via SHELLOPTS/PS4 can bypass allowlist intent (RCE) ✓ Mitigated
ASH-OCLAW-371 — MEDIUM Hook auth rate limiter bypass via IPv4-mapped IPv6 client key variants ✓ Mitigated
ASH-OCLAW-372 — MEDIUM Agent avatar symlink traversal in gateway session metadata ✓ Mitigated
ASH-OCLAW-373 — MEDIUM Hook transform path containment missed symlink-resolved escapes ✓ Mitigated
ASH-OCLAW-374 — MEDIUM macOS optional allowlist basename matching could bypass path-based policy ✓ Mitigated
ASH-OCLAW-375 — MEDIUM Exec allowlist wrapper analysis did not unwrap env/shell dispatch chains ✓ Mitigated
ASH-OCLAW-376 — MEDIUM macOS `system.run` allowlist bypass via quoted command substitution ✓ Mitigated
ASH-OCLAW-377 — MEDIUM Inbound media downloads could exceed configured byte limits before rejection across multip ✓ Mitigated
ASH-OCLAW-378 — LOW Control UI Static File Handler Follows Symlinks and Allows Out-of-Root File Read ✓ Mitigated
ASH-OCLAW-379 — MEDIUM Shell env fallback trusts unvalidated SHELL path from host environment ✓ Mitigated
ASH-OCLAW-380 — MEDIUM Node role device-identity bypass allows unauthorized node.event injection ✓ Mitigated
ASH-OCLAW-381 — MEDIUM BlueBubbles (optional plugin) pairing/allowlist mismatch when allowFrom is empty ✓ Mitigated
ASH-OCLAW-382 — MEDIUM Discord `allowFrom` slug-collision authorization bypass ✓ Mitigated
ASH-OCLAW-383 — MEDIUM Zip extraction symlink traversal could write outside destination ✓ Mitigated
ASH-OCLAW-384 — MEDIUM non-default configuration: manually adding sort to tools.exec.safeBins could bypass allowl ✓ Mitigated
ASH-OCLAW-385 — LOW macOS companion app (beta): allowlist parsing mismatch for system.run shell chains ✓ Mitigated
ASH-OCLAW-386 — MEDIUM Loopback CDP probe can leak Gateway token to local listener ✓ Mitigated
ASH-OCLAW-387 — MEDIUM DUPLICATE of GHSA-3c6h-g97w-fg78: safeBins denied flags can be bypassed via GNU long-optio ✓ Mitigated
ASH-OCLAW-388 — HIGH exec allowlist/safeBins policy-runtime mismatch via env -S wrapper interpretation ✓ Mitigated
ASH-OCLAW-389 — MEDIUM exec allow-always can be bypassed via unrecognized multiplexer shell wrappers (busybox/toy ✓ Mitigated
ASH-OCLAW-390 — MEDIUM commands.allowFrom sender authorization accepted conversation identifiers via ctx.From ✓ Mitigated
ASH-OCLAW-391 — MEDIUM image tool bypasses tools.fs.workspaceOnly on sandbox mount paths and exfiltrates out-of-w ✓ Mitigated
ASH-OCLAW-392 — MEDIUM allowlist exec-guard bypass via env -S ✓ Mitigated
ASH-OCLAW-393 — MEDIUM Node exec approvals could be replayed across nodes ✓ Mitigated
ASH-OCLAW-394 — LOW iOS deep link (openclaw://agent) can trigger gateway agent requests without local confirma ✓ Mitigated
ASH-OCLAW-395 — HIGH tools.exec.safeBins sort long-option abbreviation bypass can skip exec approval in allowli ✓ Mitigated
ASH-OCLAW-396 — MEDIUM non-default autoAllowSkills setting could bypass on-miss exec prompt ✓ Mitigated
ASH-OCLAW-397 — MEDIUM shell-env trusted-prefix fallback allowed attacker-controlled binary execution via $SHELL ✓ Mitigated
ASH-OCLAW-398 — MEDIUM Stored XSS in exported session HTML viewer via markdown/raw-HTML rendering ✓ Mitigated
ASH-OCLAW-399 — MEDIUM Zalo group sender allowlist bypass permits unauthorized GROUP dispatch ✓ Mitigated
ASH-OCLAW-400 — MEDIUM Temporary path handling could write outside OpenClaw temp boundary ✓ Mitigated
ASH-OCLAW-401 — MEDIUM safeBins static default trusted dirs allow writable-dir binary hijack (`jq`) ✓ Mitigated
ASH-OCLAW-402 — MEDIUM system.run shell-wrapper positional argv carriers could execute hidden commands under misl ✓ Mitigated
ASH-OCLAW-403 — MEDIUM Sandbox bind validation could bypass allowed-root and blocked-path checks via symlink-pare ✓ Mitigated
ASH-OCLAW-404 — MEDIUM Telegram: Unauthorized Senders Trigger Media Download and Disk Write Before Access Check ✓ Mitigated
ASH-OCLAW-405 — MEDIUM Dispatch-wrapper depth-cap mismatch can bypass shell-wrapper approval gating in system.run ✓ Mitigated
ASH-OCLAW-406 — LOW sandbox network isolation bypass via docker.network=container: ✓ Mitigated
ASH-OCLAW-407 — HIGH Native prompt image auto-load did not honor tools.fs.workspaceOnly in sandboxed runs ✓ Mitigated
ASH-OCLAW-408 — LOW IPv6 multicast SSRF classifier bypass ✓ Mitigated
ASH-OCLAW-409 — LOW macOS beta onboarding exposed PKCE verifier via OAuth state ✓ Mitigated
ASH-OCLAW-410 — MEDIUM Gateway agents.files symlink escape allowed out-of-workspace file read/write ✓ Mitigated
ASH-OCLAW-411 — MEDIUM Signal reaction-only status events could, in limited cases, be enqueued before access chec ✓ Mitigated
ASH-OCLAW-412 — MEDIUM Telegram message_reaction authorization bypass allows unauthorized system-event injection ✓ Mitigated
ASH-OCLAW-413 — MEDIUM Browser trace/download path symlink escape in temp output handling ✓ Mitigated
ASH-OCLAW-414 — MEDIUM Discord DM reaction ingress missed dmPolicy/allowFrom checks in restricted setups ✓ Mitigated
ASH-OCLAW-415 — MEDIUM Trusted-proxy Control UI pairing bypass allowed unpaired node sessions ✓ Mitigated
ASH-OCLAW-416 — HIGH Hardlink alias checks could bypass workspace-only file boundaries in specific configuratio ✓ Mitigated
ASH-OCLAW-417 — MEDIUM system.run approval TOCTOU via mutable symlink cwd target on node host ✓ Mitigated
ASH-OCLAW-418 — MEDIUM Sandbox media fallback tmp symlink alias bypass allows host file reads outside sandboxRoot ✓ Mitigated
ASH-OCLAW-419 — MEDIUM MS Teams fileConsent/invoke missing conversation binding allowed cross-conversation pendin ✓ Mitigated
ASH-OCLAW-420 — HIGH Browser-origin WebSocket auth hardening gap could enable loopback password brute-force cha ✓ Mitigated
ASH-OCLAW-421 — MEDIUM Node system.run approval bypass via parent-symlink cwd rebind ✓ Mitigated
ASH-OCLAW-422 — LOW voice-call Twilio replay dedupe now bound to authenticated webhook identity ✓ Mitigated
ASH-OCLAW-423 — LOW Cross-account DM pairing authorization bypass via unscoped pairing store access ✓ Mitigated
ASH-OCLAW-424 — LOW BlueBubbles group allowlist mismatch via DM pairing-store fallback ✓ Mitigated
ASH-OCLAW-425 — MEDIUM Microsoft Teams media fetch SSRF hardening: unified guarded fetch across Graph and attachm ✓ Mitigated
ASH-OCLAW-426 — MEDIUM Node system.run approval context-binding weakness in approval-enabled host=node flows ✓ Mitigated
ASH-OCLAW-427 — MEDIUM DM pairing-store identities could satisfy group allowlist authorization ✓ Mitigated
ASH-OCLAW-428 — HIGH Gateway plugin auth bypass via encoded dot-segment traversal in protected /api/channels pa ✓ Mitigated
ASH-OCLAW-429 — LOW Signal group allowlist authorization bypass via DM pairing-store leakage ✓ Mitigated
ASH-OCLAW-430 — HIGH LINE group allowlist scope mismatch with DM pairing-store entries ✓ Mitigated
ASH-OCLAW-431 — HIGH workspace path guard bypass on non-existent out-of-root symlink leaf ✓ Mitigated
ASH-OCLAW-432 — HIGH Sandbox dangling-symlink alias handling could bypass workspace-only write boundary ✓ Mitigated
ASH-OCLAW-433 — MEDIUM /api/channels gateway-auth boundary bypass via path canonicalization mismatch ✓ Mitigated
ASH-OCLAW-434 — MEDIUM Browser control startup could continue unauthenticated after auth bootstrap failure ✓ Mitigated
ASH-OCLAW-435 — HIGH Authorization mismatch allowed write-scope agent runs to reach owner-only tools ✓ Mitigated
ASH-OCLAW-436 — MEDIUM Unicode canonicalization drift in node metadata policy classification could broaden node a ✓ Mitigated
ASH-OCLAW-437 — HIGH system.run approvals did not bind PATH-token executable identity, enabling post-approval e ✓ Mitigated
ASH-OCLAW-438 — MEDIUM Sandbox media TOCTOU could read files outside sandbox root ✓ Mitigated
ASH-OCLAW-439 — MEDIUM TOCTOU symlink race in writeFileWithinRoot could create or truncate files outside root bou ✓ Mitigated
ASH-OCLAW-440 — HIGH ACPX Windows wrapper shell fallback allowed cwd injection in specific paths ✓ Mitigated
ASH-OCLAW-441 — MEDIUM Unauthorized sender bypass in stop triggers and /models command authorization ✓ Mitigated
ASH-OCLAW-442 — MEDIUM web_search citation redirect SSRF via private-network-allowing policy ✓ Mitigated
ASH-OCLAW-443 — MEDIUM Node camera URL payload host-binding bypass allowed gateway fetch pivots ✓ Mitigated
ASH-OCLAW-444 — MEDIUM ZIP extraction race could write outside destination via parent symlink rebind ✓ Mitigated
ASH-OCLAW-445 — MEDIUM stageSandboxMedia destination symlink traversal can overwrite files outside sandbox worksp ✓ Mitigated
ASH-OCLAW-446 — MEDIUM Node system.run approval hardening wrapper semantic drift can execute unintended local scr ✓ Mitigated
ASH-OCLAW-447 — HIGH Sandboxed sessions_spawn(runtime="acp") bypassed sandbox inheritance and allowed host ACP ✓ Mitigated
ASH-OCLAW-448 — MEDIUM Unified root-bound write hardening for browser output and related path-boundary flows ✓ Mitigated
ASH-OCLAW-449 — MEDIUM Pre-auth webhook body parsing can enable unauthenticated slow-request DoS ✓ Mitigated
ASH-OCLAW-450 — MEDIUM hooks count non-POST requests toward auth lockout ✓ Mitigated
ASH-OCLAW-451 — HIGH Dashboard leaked gateway auth material via browser URL/query and localStorage ✓ Mitigated
ASH-OCLAW-452 — HIGH fetch-guard forwards custom authorization headers across cross-origin redirects ✓ Mitigated
ASH-OCLAW-453 — MEDIUM Cross-account sender authorization expansion in `/allowlist ... --store` account scoping ✓ Mitigated
ASH-OCLAW-454 — MEDIUM `operator.write` chat.send could reach admin-only config writes ✓ Mitigated
ASH-OCLAW-455 — MEDIUM `system.run` env override filtering allowed dangerous helper-command pivots ✓ Mitigated
ASH-OCLAW-456 — MEDIUM system.run allow-always persistence included shell-commented payload tails ✓ Mitigated
ASH-OCLAW-457 — MEDIUM Sandboxed /acp spawn requests could initialize host ACP sessions ✓ Mitigated
ASH-OCLAW-458 — MEDIUM system.run allowlist approval parsing missed PowerShell encoded-command wrappers ✓ Mitigated
ASH-OCLAW-459 — MEDIUM system.run wrapper-depth boundary could skip shell approval gating ✓ Mitigated
ASH-OCLAW-460 — MEDIUM system.run approvals did not bind mutable script operands across approval and execution ✓ Mitigated
ASH-OCLAW-461 — MEDIUM skills-install-download can be redirected outside the tools root by rebinding the validate ✓ Mitigated
ASH-OCLAW-462 — MEDIUM MS Teams sender allowlist bypass when route allowlist is configured and sender allowlist i ✓ Mitigated
ASH-OCLAW-463 — CRITICAL Pairing-scoped device tokens could mint `operator.admin` and reach node RCE ✓ Mitigated
ASH-OCLAW-464 — HIGH Leaf subagents could steer sibling sessions across sandbox boundaries ✓ Mitigated
ASH-OCLAW-465 — MEDIUM Channel commands could bypass account-scoped `configWrites` restrictions ✓ Mitigated
ASH-OCLAW-466 — MEDIUM Discord guild reaction ingress could bypass users and roles allowlists ✓ Mitigated
ASH-OCLAW-467 — MEDIUM Write-scoped callers could reach admin-only session reset logic through `agent` ✓ Mitigated
ASH-OCLAW-468 — HIGH Unbound interpreter and runtime commands could bypass node-host approval integrity ✓ Mitigated
ASH-OCLAW-469 — HIGH Plugin subagent routes could bypass gateway authorization with synthetic admin scopes ✓ Mitigated
ASH-OCLAW-470 — MEDIUM Zalouser allowlist authorization matched mutable group names by default ✓ Mitigated
ASH-OCLAW-471 — CRITICAL WebSocket shared-auth connections could self-declare elevated scopes ✓ Mitigated
ASH-OCLAW-472 — HIGH Command-authorized non-owners could reach owner-only `/config` and `/debug` surfaces ✓ Mitigated
ASH-OCLAW-473 — MEDIUM Zalo webhook rate limiting could be bypassed before secret validation ✓ Mitigated
ASH-OCLAW-474 — HIGH `browser.request` let `operator.write` persist admin-only browser profile changes ✓ Mitigated
ASH-OCLAW-475 — HIGH `session_status` let sandboxed subagents access parent or sibling session state ✓ Mitigated
ASH-OCLAW-476 — HIGH Telegram webhook request bodies were read before secret validation, enabling unauthenticat ✓ Mitigated
ASH-OCLAW-477 — HIGH Bootstrap setup codes could be replayed to escalate pending pairing scopes before approval ✓ Mitigated
ASH-OCLAW-478 — MEDIUM Unsanitized iMessage attachment paths allowed SCP remote-path command injection ✓ Mitigated
ASH-OCLAW-479 — MEDIUM Session transcript files were created without forced user-only permissions ✓ Mitigated
ASH-OCLAW-480 — HIGH Gateway agent /reset exposes admin session reset to operator.write callers ✓ Mitigated
ASH-OCLAW-481 — HIGH Plivo V2 verified replay identity drifts on query-only variants ✓ Mitigated
ASH-OCLAW-482 — MEDIUM Gateway Canvas local-direct requests bypass Canvas HTTP and WebSocket authentication ✓ Mitigated
ASH-OCLAW-483 — HIGH system.run allowlist bypass via unregistered time dispatch wrapper. ✓ Mitigated
ASH-OCLAW-484 — MEDIUM Remote media error responses could trigger unbounded memory allocation before failure. ✓ Mitigated
ASH-OCLAW-485 — HIGH Synology Chat reply delivery could be rebound through username-based user resolution. ✓ Mitigated
ASH-OCLAW-486 — MEDIUM TXT-only Bonjour and DNS-SD discovery metadata could still steer CLI routing when service ✓ Mitigated
ASH-OCLAW-487 — HIGH Nostr inbound DMs could trigger unauthenticated crypto work before sender policy enforceme ✓ Mitigated
ASH-OCLAW-488 — HIGH Trusted-proxy Control UI sessions could retain self-declared privileged scopes without dev ✓ Mitigated
ASH-OCLAW-489 — MEDIUM Voice Call webhook buffered request bodies before provider signature checks, enabling boun ✓ Mitigated
ASH-OCLAW-490 — HIGH Gateway host exec env override handling did not consistently apply the shared host env pol ✓ Mitigated
ASH-OCLAW-491 — MEDIUM Sessions send Action Missing controlScope Enforcement Allows Leaf Subagents to Message Con ✓ Mitigated
ASH-OCLAW-492 — CRITICAL Gateway device.pair.approve Lets operator.pairing Escalate a New Device into operator.admi ✓ Mitigated
ASH-OCLAW-493 — MEDIUM Mattermost callback dispatch allowed non-allowlisted sender actions. ✓ Mitigated
ASH-OCLAW-494 — HIGH Google Chat app-url webhook auth accepted non-deployment add-on principals. ✓ Mitigated
ASH-OCLAW-495 — MEDIUM Tlon settings empty-allowlist reconciliation bypassed intended revocation. ✓ Mitigated
ASH-OCLAW-496 — HIGH Mutating internal ACP chat commands missed operator.admin scope enforcement. ✓ Mitigated
ASH-OCLAW-497 — MEDIUM Tlon cite expansion happened before channel and DM authorization completed. ✓ Mitigated
ASH-OCLAW-498 — MEDIUM Image tool bypassed tools.fs.workspaceOnly and could read mounted files outside the worksp ✓ Mitigated
ASH-OCLAW-499 — HIGH Gateway HTTP /sessions/:sessionKey/kill Reaches Admin Kill Path Without Caller Scope Bindi ✓ Mitigated
ASH-OCLAW-500 — MEDIUM Gateway HTTP Session History Route Bypasses Operator Read Scope ✓ Mitigated
ASH-OCLAW-501 — CRITICAL Gateway Backend Reconnect lets Non-Admin Operator Scopes Self-Claim operator.admin ✓ Mitigated
ASH-OCLAW-502 — CRITICAL Gateway local shared-auth reconnect silently widens paired device scope from operator.read ✓ Mitigated
ASH-OCLAW-503 — HIGH Gateway Plugin HTTP auth: "gateway" Mints operator.admin Runtime Scope ✓ Mitigated
ASH-OCLAW-504 — MEDIUM Synology Chat Webhook Pre-Auth Rate-Limit Bypass Enables Brute-Force Guessing of Weak Webh ✓ Mitigated
ASH-OCLAW-505 — MEDIUM Telegram Webhook Missing Guess Rate Limiting Enables Brute-Force Guessing of Weak Webhook ✓ Mitigated
ASH-OCLAW-506 — MEDIUM BlueBubbles Webhook Missing Guess Rate Limiting Enables Brute-Force Guessing of Weak Webho ✓ Mitigated
ASH-OCLAW-507 — MEDIUM Google Chat Authz Bypass via Group Policy Rebinding with Mutable Space displayName ✓ Mitigated
ASH-OCLAW-508 — MEDIUM Matrix Verification Notices Bypass Matrix DM Policy and Reply to Unpaired DM Peers ✓ Mitigated
ASH-OCLAW-509 — MEDIUM Telegram DM-Scoped Inline Button Callbacks Bypass DM Pairing and Mutate Session State ✓ Mitigated
ASH-OCLAW-510 — HIGH `session_status` sessionId resolution bypasses sandboxed session-tree visibility ✓ Mitigated
ASH-OCLAW-511 — MEDIUM Feishu webhook reads and parses unauthenticated request bodies before signature validation ✓ Mitigated
ASH-OCLAW-512 — HIGH SSRF via Unguarded Configured Base URLs in Multiple Channel Extensions (Incomplete Fix for ✓ Mitigated
ASH-OCLAW-513 — MEDIUM Gateway HTTP /v1/models Route Bypasses Operator Read Scope ✓ Mitigated
ASH-OCLAW-514 — HIGH Gateway operator.write Can Reach Admin-Class Channel Allowlist Persistence via chat.send ✓ Mitigated
ASH-OCLAW-515 — MEDIUM Mutating internal `/allowlist` chat commands missed `operator.admin` scope enforcement ✓ Mitigated
ASH-OCLAW-516 — MEDIUM Incomplete Fix for CVE-2026-27486: Unvalidated SIGKILL in `!stop` Chat Command via `shell- ✓ Mitigated
ASH-OCLAW-517 — HIGH `browser.request` still allows `POST /reset-profile` through the `operator.write` surface ✓ Mitigated
ASH-OCLAW-518 — HIGH Voice-call Plivo V3 webhook replay key uses unsorted URL, allowing replay via query-parame ✓ Mitigated
ASH-OCLAW-519 — MEDIUM Zalo channel downloads media before sender authorization ✓ Mitigated
ASH-OCLAW-520 — HIGH CLI Remote Onboarding Persists Unauthenticated Discovery Endpoint and Exfiltrates Gateway ✓ Mitigated
ASH-OCLAW-521 — HIGH node.pair.approve missing callerScopes validation allows low-privilege operator to approve ✓ Mitigated
ASH-OCLAW-522 — HIGH Discord text `/approve` bypasses `channels.discord.execApprovals.approvers` and allows non ✓ Mitigated
ASH-OCLAW-523 — MEDIUM Discord component interaction ingress skips guild/channel policy enforcement ✓ Mitigated
ASH-OCLAW-524 — HIGH Device removal and token revocation do not terminate active WebSocket sessions ✓ Mitigated
ASH-OCLAW-525 — HIGH Gateway exec allow-always over-trusts positional carrier executables ✓ Mitigated
ASH-OCLAW-526 — MEDIUM OpenClaw host-env blocklist missing `GIT_TEMPLATE_DIR` and `AWS_CONFIG_FILE` allows code e ✓ Mitigated
ASH-OCLAW-527 — HIGH message tool media parameter bypasses tool policy filesystem isolation ✓ Mitigated
ASH-OCLAW-528 — HIGH Gateway `operator.write` can reach admin-only session reset via `chat.send` `/reset` ✓ Mitigated
ASH-OCLAW-529 — LOW OpenClaw SSRF guard misses four IPv6 special-use ranges ✓ Mitigated
ASH-OCLAW-530 — MEDIUM Gateway `operator.write` can reach admin-only persisted `verboseLevel` via `chat.send` `/v ✓ Mitigated
ASH-OCLAW-531 — HIGH OpenClaw Gateway WebSocket denial of service via unbounded pre-auth upgrades ✓ Mitigated
ASH-OCLAW-532 — HIGH OpenClaw safeBins jq `$ENV` filter bypass allows environment variable disclosure ✓ Mitigated
ASH-OCLAW-533 — HIGH Gateway chat.send ACP-only provenance guard could be bypassed by client identity spoofing ✓ Mitigated
ASH-OCLAW-534 — HIGH Gateway exec allowlist allow-always bypass via unregistered /usr/bin/script wrapper ✓ Mitigated
ASH-OCLAW-535 — MEDIUM Feishu extension resolveUploadInput bypasses file-system sandbox and allows arbitrary file ✓ Mitigated
ASH-OCLAW-536 — CRITICAL CWD `.env` environment variable injection bypasses host-env policy and allows config takeo ✓ Mitigated
ASH-OCLAW-537 — CRITICAL Incomplete host env sanitization blocklist allows supply-chain redirection via package-man ✓ Mitigated
ASH-OCLAW-538 — MEDIUM Nextcloud Talk webhook missing rate limiting on shared secret authentication ✓ Mitigated
ASH-OCLAW-539 — MEDIUM Unauthenticated plugin-auth HTTP routes receive operator runtime scopes ✓ Mitigated
ASH-OCLAW-540 — MEDIUM OpenClaw Nostr privateKey config redaction bypass leaks plaintext signing key via config.g ✓ Mitigated
ASH-OCLAW-541 — LOW Voice-call Plivo replay mutates in-process callback origin before replay rejection ✓ Mitigated
ASH-OCLAW-542 — HIGH Workspace `.env` can override the bundled hooks root and load attacker hook code ✓ Mitigated
ASH-OCLAW-543 — LOW Zalo webhook replay cache cross-target messageId scope bypass ✓ Mitigated
ASH-OCLAW-544 — HIGH Workspace `.env` can override the bundled plugin trust root ✓ Mitigated
ASH-OCLAW-545 — LOW Gateway `device.token.rotate` does not terminate active WebSocket sessions after credentia ✓ Mitigated
ASH-OCLAW-546 — CRITICAL Sandbox escape via TOCTOU race in remote FS bridge readFile ✓ Mitigated
ASH-OCLAW-547 — HIGH `PIP_INDEX_URL` and `UV_INDEX_URL` bypass host exec env sanitization and redirect Python p ✓ Mitigated
ASH-OCLAW-548 — MEDIUM Incomplete fix for CVE-2026-32062: voice-call still parses large WebSocket frames before s ✓ Mitigated
ASH-OCLAW-549 — MEDIUM Discord voice ingress authorization can be bypassed via channel, name, and stale-role vali ✓ Mitigated
ASH-OCLAW-550 — MEDIUM Feishu thread history and quoted messages bypass sender allowlist ✓ Mitigated
ASH-OCLAW-551 — MEDIUM Matrix thread root and reply context bypass sender allowlist ✓ Mitigated
ASH-OCLAW-552 — LOW Fake DeviceToken Bypasses Shared Auth Rate Limiting ✓ Mitigated
ASH-OCLAW-553 — CRITICAL Heartbeat context inheritance bypasses sandbox via senderIsOwner escalation ✓ Mitigated
ASH-OCLAW-554 — LOW Security Scan Failure Does Not Block Plugin Installation (Fail-Open) ✓ Mitigated
ASH-OCLAW-555 — MEDIUM MSTeams thread history bypasses sender allowlist via Graph API ✓ Mitigated
ASH-OCLAW-556 — HIGH Device-Paired Node Skips Node Scope Gate → Host RCE.md ✓ Mitigated
ASH-OCLAW-557 — LOW LINE webhook handler lacks shared pre-auth concurrency budget before signature verificatio ✓ Mitigated
ASH-OCLAW-558 — LOW Telnyx Webhook Replay Detection Bypass via Base64 Signature Re-encoding ✓ Mitigated
ASH-OCLAW-559 — LOW Tlon media downloads can bypass core safety limits and exhaust disk ✓ Mitigated
ASH-OCLAW-560 — MEDIUM SSRF via Unguarded `fetch()` in Marketplace Plugin Download and Ollama Model Discovery ✓ Mitigated
ASH-OCLAW-561 — MEDIUM Incomplete `host-env-security-policy.json` allows untrusted model to substitute compiler b ✓ Mitigated
ASH-OCLAW-562 — MEDIUM Image pixel-limit guard can fail open on sips and allow decompression-bomb DoS ✓ Mitigated
ASH-OCLAW-563 — LOW SSH-based sandbox backends pass unsanitized process.env to child processes ✓ Mitigated
ASH-OCLAW-564 — HIGH SSH sandbox tar upload follows symlinks, enabling arbitrary file write on remote host ✓ Mitigated
ASH-OCLAW-565 — MEDIUM Media download follows cross-origin redirects with Authorization headers intact ✓ Mitigated
ASH-OCLAW-566 — MEDIUM Marketplace Plugin Download Follows Redirects Without SSRF Protection ✓ Mitigated
ASH-OCLAW-567 — MEDIUM Host exec environment overrides miss proxy, TLS, Docker, and Git TLS controls ✓ Mitigated
ASH-OCLAW-568 — MEDIUM Media Local Roots Self-Whitelisting in appendLocalMediaParentRoots Allows Mod ✓ Mitigated
ASH-OCLAW-569 — MEDIUM Host exec environment sanitization misses package, registry, Docker, compiler, and TLS ove ✓ Mitigated
ASH-OCLAW-570 — MEDIUM Telegram audio preflight transcription enables resource consumption by unauthorized sender ✓ Mitigated
ASH-OCLAW-571 — MEDIUM Discord audio preflight transcription before member authorization ✓ Mitigated
ASH-OCLAW-572 — HIGH OpenShell Mirror Sync: Sandbox Escape via Unrestricted File Sync + Symlink Traversal ✓ Mitigated
ASH-OCLAW-573 — HIGH Paired node escalates to gateway RCE via unrestricted node.event agent dispatch ✓ Mitigated
ASH-OCLAW-574 — MEDIUM MS Teams webhook parses body before JWT validation, enabling unauthenticated resource exha ✓ Mitigated
ASH-OCLAW-575 — MEDIUM HTTP operator endpoints lack browser-origin validation in trusted-proxy mode ✓ Mitigated
ASH-OCLAW-576 — HIGH Incomplete scope-clearing fix allows operator.admin escalation via trusted-proxy auth mode ✓ Mitigated
ASH-OCLAW-577 — MEDIUM Path traversal via inbound channel attachment path in ACP dispatch allows arbitrary file r ✓ Mitigated
ASH-OCLAW-578 — MEDIUM Sandbox file operations use check-then-act, bypassing fd-based TOCTOU defenses ✓ Mitigated
ASH-OCLAW-579 — LOW OpenClaw Gateway Control Interface Information Disclosure Vulnerability ✓ Mitigated
ASH-OCLAW-580 — MEDIUM `/phone arm`/`/phone disarm` Bypasses `operator.admin` Scope Check for External Channels ✓ Mitigated
ASH-OCLAW-581 — HIGH Agentic Consent Bypass: LLM Agent Can Silently Disable Exec Approval via `config.patch` ✓ Mitigated
ASH-OCLAW-582 — HIGH OpenClaw <= 2026.3.23 Media Parsing Path Traversal to Arbitrary File Read ✓ Mitigated
ASH-OCLAW-583 — HIGH Incomplete Fix for CVE-2026-4039: CLI Backend Environment Variable Injection via Workspace ✓ Mitigated
ASH-OCLAW-584 — MEDIUM `session_status` still bypasses configured `tools.sessions.visibility` for unsandboxed inv ✓ Mitigated
ASH-OCLAW-585 — MEDIUM Gateway operator.write Can Reach Admin-Class Telegram Config and Cron Persistence via send ✓ Mitigated
ASH-OCLAW-586 — MEDIUM Gateway operator.write Can Reach Admin-Class Talk Voice Config Persistence via chat.send ✓ Mitigated
ASH-OCLAW-587 — MEDIUM OpenShell `mirror` mode can convert untrusted sandbox files into explicitly enabled worksp ✓ Mitigated
ASH-OCLAW-588 — MEDIUM Discord voice manager bypasses channel-level member access allowlist ✓ Mitigated
ASH-OCLAW-589 — LOW Tlon Startup Migration Rehydrates Empty-Array Revocations From File Config ✓ Mitigated
ASH-OCLAW-590 — LOW diffs viewer misclassifies proxied remote requests as loopback when `allowRemoteViewer` is ✓ Mitigated
ASH-OCLAW-591 — LOW Telegram legacy allowFrom migration fans default-account trust into all named accounts ✓ Mitigated
ASH-OCLAW-592 — MEDIUM macOS Wide-Area Discovery Accepts Arbitrary Tailnet Peer as DNS Authority and Exfiltrates ✓ Mitigated
ASH-OCLAW-593 — MEDIUM Remote onboarding preserves attacker-discovered endpoint after trust decline, routing gate ✓ Mitigated
ASH-OCLAW-594 — LOW Discord Slash Commands Bypass Group DM Channel Allowlist ✓ Mitigated
ASH-OCLAW-595 — LOW Discord Component Interaction Misclassifies Group DM as Direct Message ✓ Mitigated
ASH-OCLAW-596 — HIGH Node browser proxy `allowProfiles` bypass through persistent profile mutation and runtime ✓ Mitigated
ASH-OCLAW-597 — HIGH Bootstrap setup codes were not bound to the intended device role and scopes, allowing firs ✓ Mitigated
ASH-OCLAW-598 — MEDIUM Forged Nostr DMs could create pairing state before signature verification ✓ Mitigated
ASH-OCLAW-599 — MEDIUM Shell init-file options could satisfy exec allowlist script matching ✓ Mitigated
ASH-OCLAW-600 — LOW Zalo replay dedupe cache could suppress events across authenticated webhook targets ✓ Mitigated
ASH-OCLAW-601 — LOW Pairing pending-request caps were enforced per channel instead of per account ✓ Mitigated
ASH-OCLAW-602 — MEDIUM Untrusted workspace channel shadows could execute during built-in channel setup ✓ Mitigated
ASH-OCLAW-603 — MEDIUM Read-scoped identity-bearing HTTP clients could kill sessions via /sessions/:sessionKey/ki ✓ Mitigated
ASH-OCLAW-604 — MEDIUM Windows-compatible env override keys could bypass system.run approval binding ✓ Mitigated
ASH-OCLAW-605 — MEDIUM Complex interpreter pipelines could skip exec script preflight validation ✓ Mitigated
ASH-OCLAW-606 — HIGH QQ Bot structured payloads could read arbitrary local files ✓ Mitigated
ASH-OCLAW-607 — MEDIUM iOS A2UI bridge trusted generic local-network pages for agent.request dispatch ✓ Mitigated
ASH-OCLAW-608 — HIGH OpenShell mirror mode could delete arbitrary remote directories when roots were mis-scoped ✓ Mitigated
ASH-OCLAW-609 — LOW Gateway hello snapshots exposed host config and state paths to non-admin clients ✓ Mitigated
ASH-OCLAW-610 — LOW Shared-secret comparison call sites leaked length information through timing ✓ Mitigated
ASH-OCLAW-611 — MEDIUM Zalo replay dedupe keys could suppress messages across chats or senders ✓ Mitigated
ASH-OCLAW-612 — HIGH pnpm dlx approvals did not bind local script operands ✓ Mitigated
ASH-OCLAW-613 — MEDIUM Slack thread context could include messages from non-allowlisted senders ✓ Mitigated
ASH-OCLAW-614 — LOW Android accepted cleartext remote gateway endpoints and sent stored credentials over ws:// ✓ Mitigated
ASH-OCLAW-615 — LOW Trailing-dot localhost CDP hosts could bypass remote loopback protections ✓ Mitigated
ASH-OCLAW-616 — MEDIUM Feishu docx upload_file/upload_image Bypasses Workspace-Only Filesystem Policy (GHSA-qf48- ✓ Mitigated
ASH-OCLAW-617 — HIGH Node Pairing Reconnect Command Escalation Bypasses operator.admin Scope Requirement ✓ Mitigated
ASH-OCLAW-618 — HIGH Lower-trust background runtime output is injected into trusted `System:` events, and local ✓ Mitigated
ASH-OCLAW-619 — HIGH Shared reply MEDIA: paths are treated as trusted and can trigger cross-channel local file ✓ Mitigated
ASH-OCLAW-620 — MEDIUM QQ Bot Extension: Missing SSRF Protection on All Media Fetch Paths ✓ Mitigated
ASH-OCLAW-621 — HIGH Authenticated `/hooks/wake` and mapped `wake` payloads are promoted into the trusted `Syst ✓ Mitigated
ASH-OCLAW-622 — MEDIUM Browser SSRF Policy Bypass via Interaction-Triggered Navigation ✓ Mitigated
ASH-OCLAW-623 — MEDIUM Multiple Code Paths Missing Base64 Pre-Allocation Size Checks ✓ Mitigated
ASH-OCLAW-624 — HIGH strictInlineEval explicit-approval boundary bypassed by approval-timeout fallback on gatew ✓ Mitigated
ASH-OCLAW-625 — HIGH OpenClaw `node.pair.approve` placed in `operator.write` scope instead of `operator.pairing ✓ Mitigated
ASH-OCLAW-626 — HIGH HGRCPATH, CARGO_BUILD_RUSTC_WRAPPER, RUSTC_WRAPPER, and MAKEFLAGS missing from exec env de ✓ Mitigated
ASH-OCLAW-627 — HIGH OpenClaw `node.invoke(browser.proxy)` bypasses `browser.request` persistent profile-mutati ✓ Mitigated
ASH-OCLAW-628 — HIGH OpenClaw `device.token.rotate` mints tokens for unapproved roles, bypassing device role-up ✓ Mitigated
ASH-OCLAW-629 — HIGH GIT_DIR and related git plumbing env vars missing from exec env denylist (GHSA-m866-6qv5-p ✓ Mitigated
ASH-OCLAW-630 — MEDIUM Existing WS sessions survive shared gateway token rotation ✓ Mitigated
ASH-OCLAW-631 — MEDIUM resolvedAuth closure becomes stale after config reload ✓ Mitigated
ASH-OCLAW-632 — HIGH B-M3: ClawHub package downloads are not enforced with integrity verification ✓ Mitigated
ASH-OCLAW-633 — MEDIUM /allowlist omits owner-only enforcement for cross-channel allowlist writes ✓ Mitigated
ASH-OCLAW-634 — HIGH Gateway plugin HTTP `auth: gateway` widens identity-bearing `operator.read` requests into ✓ Mitigated
ASH-OCLAW-635 — MEDIUM Concurrent async auth attempts can bypass the intended shared-secret rate-limit budget on ✓ Mitigated
ASH-OCLAW-636 — HIGH `fetchWithSsrFGuard` replays unsafe request bodies across cross-origin redirects ✓ Mitigated
ASH-OCLAW-637 — MEDIUM Strict browser SSRF bypass in Playwright redirect handling leaves private targets reachabl ✓ Mitigated
ASH-OCLAW-638 — HIGH OpenClaw Host-Exec Environment Variable Injection ✓ Mitigated
ASH-OCLAW-639 — LOW TOCTOU read in exec script preflight ✓ Mitigated
ASH-OCLAW-640 — MEDIUM Browser snapshot and screenshot routes could expose internal page content after navigation ✓ Mitigated
ASH-OCLAW-641 — HIGH config.get redaction bypass through sourceConfig and runtimeConfig aliases ✓ Mitigated
ASH-OCLAW-642 — LOW Microsoft Teams SSO invoke handler missed sender authorization checks ✓ Mitigated
ASH-OCLAW-643 — MEDIUM Collect-mode queue batches could reuse the last sender authorization context ✓ Mitigated
ASH-OCLAW-644 — MEDIUM Heartbeat owner downgrade missed untrusted webhook wake events ✓ Mitigated
ASH-OCLAW-645 — MEDIUM Delivery queue recovery could lose group tool-policy context for media replay ✓ Mitigated
ASH-OCLAW-646 — MEDIUM Empty approver lists could grant explicit approval authorization ✓ Mitigated
ASH-OCLAW-647 — MEDIUM Discord event cover images bypassed sandbox media normalization ✓ Mitigated
ASH-OCLAW-648 — MEDIUM Heartbeat owner downgrade missed local async exec completion events ✓ Mitigated
ASH-OCLAW-649 — MEDIUM Agent hook events could enqueue trusted system events from unsanitized external input ✓ Mitigated
ASH-OCLAW-650 — MEDIUM Voice-call realtime WebSocket accepted oversized frames ✓ Mitigated
ASH-OCLAW-651 — HIGH Sandboxed agents could escape exec routing via host=node override ✓ Mitigated
ASH-OCLAW-652 — MEDIUM Matrix profile config persistence was reachable from operator.write message tools ✓ Mitigated
ASH-OCLAW-653 — MEDIUM Memory dreaming config persistence was reachable from operator.write commands ✓ Mitigated
ASH-OCLAW-654 — MEDIUM Nostr profile mutation routes allowed operator.write config persistence ✓ Mitigated
ASH-OCLAW-655 — MEDIUM Shell-wrapper detection missed env-argv assignment injection forms ✓ Mitigated
ASH-OCLAW-656 — HIGH busybox and toybox applet execution weakened exec approval binding ✓ Mitigated
ASH-OCLAW-657 — HIGH Exec environment denylist missed high-risk interpreter startup variables ✓ Mitigated
ASH-OCLAW-658 — HIGH Workspace .env could inject OpenClaw runtime-control variables ✓ Mitigated
ASH-OCLAW-659 — HIGH Channel setup catalog lookups could include untrusted workspace plugin shadows ✓ Mitigated
ASH-OCLAW-660 — MEDIUM Remote marketplace repository paths could escape through symlink traversal ✓ Mitigated
ASH-OCLAW-661 — HIGH Workspace provider auth choices could auto-enable untrusted provider plugins ✓ Mitigated
ASH-OCLAW-662 — HIGH Browser SSRF hostname validation could be bypassed by DNS rebinding ✓ Mitigated
ASH-OCLAW-663 — HIGH Browser SSRF policy default allowed private-network navigation ✓ Mitigated
ASH-OCLAW-664 — HIGH QQBot reply media URL handling could trigger SSRF and re-upload fetched bytes ✓ Mitigated
ASH-OCLAW-665 — HIGH Sandbox browser CDP relay could expose DevTools protocol on 0.0.0.0 ✓ Mitigated
ASH-OCLAW-666 — MEDIUM screen_record outPath bypassed workspace-only filesystem guard ✓ Mitigated
ASH-OCLAW-667 — HIGH Browser interaction routes could pivot into local CDP and regain file reads ✓ Mitigated
ASH-OCLAW-668 — HIGH Browser tabs action select and close routes bypassed SSRF policy ✓ Mitigated
ASH-OCLAW-669 — MEDIUM CDP /json/version WebSocket URL could pivot to untrusted second-hop targets ✓ Mitigated
ASH-OCLAW-670 — HIGH QQBot media tags could read arbitrary local files through reply text ✓ Mitigated
ASH-OCLAW-671 — MEDIUM Existing-session browser interaction routes bypassed SSRF policy enforcement ✓ Mitigated
ASH-OCLAW-672 — HIGH Sandbox noVNC helper route exposed interactive browser session credentials ✓ Mitigated
ASH-OCLAW-673 — MEDIUM Browser press/type interaction routes missed complete navigation guard coverage ✓ Mitigated
ASH-OCLAW-674 — MEDIUM Sender policy bypass in host media attachment reads allows unauthorized local file disclos ✓ Mitigated
ASH-OCLAW-675 — HIGH Matrix room control-command authorization no longer trusts DM pairing-store entries ✓ Mitigated
ASH-OCLAW-676 — MEDIUM Webchat media embedding enforces local-root containment for tool-result files ✓ Mitigated
ASH-OCLAW-677 — HIGH Gateway HTTP endpoints re-resolve bearer auth after SecretRef rotation ✓ Mitigated
ASH-OCLAW-678 — LOW QMD memory_get restricts reads to canonical or indexed memory paths ✓ Mitigated
ASH-OCLAW-679 — CRITICAL Feishu webhook and card-action validation now fail closed ✓ Mitigated
ASH-OCLAW-680 — MEDIUM MCP stdio server env could load dangerous startup variables from workspace config ✓ Mitigated
ASH-OCLAW-681 — MEDIUM Agent gateway config mutations could change protected operator settings ✓ Mitigated
ASH-OCLAW-682 — MEDIUM Hook mapping templates could bypass hook session-key opt-in ✓ Mitigated
ASH-OCLAW-683 — LOW QQBot direct media upload skipped URL SSRF validation ✓ Mitigated
ASH-OCLAW-684 — MEDIUM Workspace dotenv could override OpenClaw runtime-control environment variables ✓ Mitigated
ASH-OCLAW-685 — LOW Paired-device pairing actions were not limited to the caller device ✓ Mitigated
ASH-OCLAW-686 — LOW Assistant media route missed scope enforcement for trusted-proxy auth ✓ Mitigated
ASH-OCLAW-687 — MEDIUM Feishu card actions could misclassify DMs and skip dmPolicy ✓ Mitigated
ASH-OCLAW-688 — LOW Isolated cron awareness events were recorded as trusted system events ✓ Mitigated
ASH-OCLAW-689 — MEDIUM Workspace dotenv MiniMax host override could redirect credentialed requests ✓ Mitigated
ASH-OCLAW-690 — LOW Browser CDP profile creation skipped strict-mode SSRF checks ✓ Mitigated
ASH-OCLAW-691 — MEDIUM Bundled MCP/LSP tools could bypass configured tool policy ✓ Mitigated
ASH-OCLAW-692 — MEDIUM Owner-enforced commands could accept wildcard channel senders as command owners ✓ Mitigated
ASH-OCLAW-693 — MEDIUM Webchat audio embedding could read local files without local-root containment ✓ Mitigated
ASH-OCLAW-694 — LOW Gateway Control UI bootstrap config required Gateway auth ✓ Mitigated
ASH-OCLAW-695 — MEDIUM Workspace dotenv files cannot override connector endpoint hosts ✓ Mitigated
ASH-OCLAW-696 — MEDIUM Exec allowlist analysis rejects shell expansion in unquoted heredocs ✓ Mitigated
ASH-OCLAW-697 — HIGH MCP loopback owner context is derived from server-issued bearer tokens ✓ Mitigated
ASH-OCLAW-698 — HIGH OpenShell FS bridge writes stay pinned to the sandbox mount root ✓ Mitigated
ASH-OCLAW-699 — HIGH OpenShell FS bridge reads pin and verify the opened file before returning bytes ✓ Mitigated
ASH-OCLAW-700 — MEDIUM ACP child sessions inherit subagent security envelope constraints ✓ Mitigated
ASH-OCLAW-701 — MEDIUM Zalo outbound photo URLs are validated through the SSRF guard ✓ Mitigated
ASH-OCLAW-702 — MEDIUM Webhooks SecretRef route secret remains valid after rotation/reload ✓ Mitigated
ASH-OCLAW-703 — HIGH Arbitrary code execution via attacker-controlled setup-api.js loaded from cwd during env-k ✓ Mitigated
ASH-OCLAW-704 — HIGH Gateway config mutation guard allowed unsafe model-driven config writes ✓ Mitigated
ASH-OCLAW-705 — MEDIUM Host environment sanitizer missed two Node.js control variables ✓ Mitigated
ASH-OCLAW-706 — MEDIUM Exec allowlist could miss side effects from transparent command wrappers ✓ Mitigated
ASH-OCLAW-707 — MEDIUM Hostname checks could treat trailing-dot hosts inconsistently ✓ Mitigated
ASH-OCLAW-708 — MEDIUM Message read actions could skip channel allowlist checks ✓ Mitigated
ASH-OCLAW-709 — MEDIUM Shell wrapper argv could change between approval and execution ✓ Mitigated
ASH-OCLAW-710 — MEDIUM Combined POSIX shell options could confuse exec revalidation ✓ Mitigated
ASH-OCLAW-711 — MEDIUM Bundle MCP loopback could miss its exec denylist on session spawn ✓ Mitigated
ASH-OCLAW-712 — HIGH Exec approval display truncation could hide the command being approved ✓ Mitigated
ASH-OCLAW-713 — MEDIUM Shell positional parameters could weaken strict inline-eval checks ✓ Mitigated
ASH-OCLAW-714 — HIGH POSIX node system.run safe-bin allowlist could be widened by shell expansion ✓ Mitigated
ASH-OCLAW-715 — MEDIUM Marketplace runtime extension metadata could point at unscanned payloads ✓ Mitigated
ASH-OCLAW-716 — HIGH Browser act interactions could bypass private-network navigation checks ✓ Mitigated
ASH-OCLAW-717 — HIGH Paired nodes could forge exec lifecycle events without system.run provenance ✓ Mitigated
ASH-OCLAW-718 — HIGH Pairing-scoped device session could restore revoked node token authority ✓ Mitigated
ASH-OCLAW-719 — HIGH Hook-triggered CLI runs could receive owner MCP tool authority ✓ Mitigated
ASH-OCLAW-720 — MEDIUM Same-host trusted-proxy deployments could accept local forged identity headers ✓ Mitigated
ASH-OCLAW-721 — HIGH Control UI locality spoofing could mint a durable admin device token ✓ Mitigated
ASH-OCLAW-722 — HIGH MCP Streamable HTTP redirects could forward configured custom headers to another origin ✓ Mitigated
ASH-OCLAW-723 — MEDIUM Bootstrap token replay could widen pending pairing scopes ✓ Mitigated
ASH-OCLAW-724 — MEDIUM Shell inline-command parsing could miss an allowlist check ✓ Mitigated
ASH-OCLAW-725 — LOW Slack reaction events could ignore reaction notification settings ✓ Mitigated
ASH-OCLAW-726 — MEDIUM Slack plugin approvals used the exec approver gate for plugin actions ✓ Mitigated
ASH-OCLAW-727 — HIGH QQBot native approval buttons did not enforce configured approver identity ✓ Mitigated
ASH-OCLAW-728 — HIGH Scoped chat.send route inheritance could bypass admin command scope gates ✓ Mitigated
ASH-OCLAW-729 — LOW BlueBubbles sender policy could match mutable conversation identifiers ✓ Mitigated
ASH-OCLAW-730 — MEDIUM Exported session HTML could keep unsafe markdown links ✓ Mitigated
ASH-OCLAW-731 — MEDIUM Node pairing reconnection could confuse approval scope state ✓ Mitigated
ASH-OCLAW-732 — MEDIUM PowerShell encoded-command aliases could miss exec allowlist checks ✓ Mitigated
ASH-OCLAW-733 — MEDIUM Trusted retry endpoint checks could match hostname prefixes ✓ Mitigated
ASH-OCLAW-734 — MEDIUM memory-wiki ingest could read local files with operator.write scope ✓ Mitigated
ASH-OCLAW-735 — MEDIUM [AgentGG] Discord allowFrom could bind to mutable display names ✓ Mitigated
ASH-OCLAW-736 — MEDIUM [AgentGG] Matrix allowFrom could bind to mutable display names ✓ Mitigated
ASH-OCLAW-737 — MEDIUM Native command authorization could skip owner-command enforcement ✓ Mitigated
ASH-OCLAW-738 — MEDIUM Telegram interactive callbacks could skip commands.allowFrom ✓ Mitigated
ASH-OCLAW-739 — MEDIUM Active Memory write scope could mutate global config ✓ Mitigated
ASH-OCLAW-740 — LOW Skill-command dispatch could skip before-tool-call hooks ✓ Mitigated
ASH-OCLAW-741 — MEDIUM Skill Workshop apply flow could override pending approval ✓ Mitigated
ASH-OCLAW-742 — LOW Feishu dynamic-agent bindings could miss configWrites enforcement ✓ Mitigated
ASH-OCLAW-743 — MEDIUM macOS Swift exec allowlist missed combined POSIX inline flags ✓ Mitigated
ASH-OCLAW-744 — MEDIUM Mattermost handlers could fall open when channel type was missing ✓ Mitigated
ASH-OCLAW-745 — MEDIUM [AgentGG] Slack allowFrom could bind to mutable display names ✓ Mitigated
ASH-OCLAW-746 — MEDIUM Config recovery could restore openclaw.json with broad file permissions ✓ Mitigated
ASH-OCLAW-747 — HIGH Trusted-proxy Control UI WebSocket accepted client-declared scopes before pairing ✓ Mitigated
ASH-OCLAW-748 — MEDIUM [AgentGG] Zalo allowFrom could bind to mutable display names ✓ Mitigated
ASH-OCLAW-749 — HIGH Non-owner chat senders could issue device-pairing bootstrap codes ✓ Mitigated
ASH-OCLAW-750 — MEDIUM QQBot admin commands could skip DM-only and allowFrom policy ✓ Mitigated
ASH-OCLAW-751 — MEDIUM QQBot streaming command could mutate config without explicit allowFrom ✓ Mitigated
ASH-OCLAW-752 — MEDIUM message.action forwarding could send Gateway credentials to model-supplied loopback URLs ✓ Mitigated
ASH-OCLAW-753 — MEDIUM Browser debug/export routes could reuse already-open blocked tabs ✓ Mitigated
ASH-OCLAW-754 — MEDIUM memory-wiki shared search could miss session visibility checks ✓ Mitigated
ASH-OCLAW-755 — HIGH Linux and macOS exec allowlists skipped configured argument patterns ✓ Mitigated
ASH-OCLAW-756 — MEDIUM QQBot pre-dispatch slash commands could skip allowFrom checks ✓ Mitigated
ASH-OCLAW-757 — MEDIUM Workspace .env could override Homebrew executable selection for skill install flows ✓ Mitigated
ASH-OCLAW-758 — MEDIUM Workspace .env CLOUDSDK_PYTHON could influence Gmail setup gcloud execution ✓ Mitigated
ASH-OCLAW-759 — MEDIUM Workspace .env STATE_DIRECTORY could influence bundled runtime dependency roots ✓ Mitigated
ASH-OCLAW-760 — MEDIUM Fake package roots could influence memory-core artifact loading ✓ Mitigated
ASH-OCLAW-761 — MEDIUM Workspace-derived service PATH could influence trash command selection ✓ Mitigated
ASH-OCLAW-762 — MEDIUM Workspace .env npm_execpath could influence bundled runtime dependency install ✓ Mitigated
ASH-OCLAW-763 — MEDIUM Sandboxed session spawn could expose the real workspace path to child prompts ✓ Mitigated
ASH-OCLAW-764 — MEDIUM Empty-scope device re-pairing could confuse caller scope containment ✓ Mitigated
ASH-OCLAW-765 — MEDIUM Tool group policy callers could accept unvalidated group IDs ✓ Mitigated
ASH-OCLAW-766 — MEDIUM Focus command could miss controlScope enforcement ✓ Mitigated
ASH-OCLAW-767 — MEDIUM Embedded runner policy could be confused by provider aliases ✓ Mitigated
ASH-OCLAW-768 — MEDIUM Internal/webchat command auth could inherit ownerAllowFrom wildcard state ✓ Mitigated
ASH-OCLAW-769 — MEDIUM Mattermost slash token revocation could lag until monitor refresh ✓ Mitigated
ASH-OCLAW-770 — MEDIUM MCP loopback could skip owner-only tool policy for non-owner callers ✓ Mitigated
ASH-OCLAW-771 — MEDIUM Slack and Zalo webhook secrets could remain active after secrets.reload ✓ Mitigated
ASH-OCLAW-772 — HIGH OpenAI-compatible HTTP model overrides could miss admin authorization ✓ Mitigated
ASH-OCLAW-773 — HIGH Feishu tools could ignore per-account disablement ✓ Mitigated
ASH-OCLAW-774 — HIGH Feishu permission tools could ignore per-account disablement ✓ Mitigated
ASH-OCLAW-775 — HIGH OpenShell mirror sync could follow remote symlink parents ✓ Mitigated
ASH-OCLAW-776 — HIGH flock wrapper could bypass durable exec approval binding ✓ Mitigated
ASH-OCLAW-777 — HIGH Message mutations could skip requester authorization ✓ Mitigated
ASH-OCLAW-778 — HIGH Discord guild actions could skip cross-provider requester authorization ✓ Mitigated
ASH-OCLAW-779 — MEDIUM Plugin install wrappers could skip install policy ✓ Mitigated
ASH-OCLAW-780 — HIGH Plugin install commands could allow non-owner persistence ✓ Mitigated
ASH-OCLAW-781 — HIGH MCP loopback could expose owner-only tools to non-owner runs ✓ Mitigated
ASH-OCLAW-782 — HIGH WhatsApp group IDs could satisfy elevated sender allowlists ✓ Mitigated
ASH-OCLAW-783 — HIGH Browser CDP discovery could accept blocked WebSocket URLs ✓ Mitigated
ASH-OCLAW-784 — MEDIUM Native web search could ignore OpenClaw tool policy ✓ Mitigated
ASH-OCLAW-785 — HIGH Host exec environment filtering could miss interpreter startup variables ✓ Mitigated
ASH-OCLAW-786 — HIGH Host exec environment filtering could allow Git ext transport ✓ Mitigated
ASH-OCLAW-787 — HIGH Sandbox exec-server HTTP requests could reach internal networks ✓ Mitigated
ASH-OCLAW-788 — HIGH Isolated cron jobs could regain denied exec tools ✓ Mitigated
ASH-OCLAW-789 — HIGH Host exec environment filtering could miss rustup startup variables ✓ Mitigated
ASH-OCLAW-790 — HIGH Sandbox bind mounts could bypass parent-directory denylist checks ✓ Mitigated
ASH-OCLAW-791 — HIGH MS Teams message actions could miss requester authorization ✓ Mitigated
ASH-OCLAW-792 — HIGH Discord moderation actions could miss trusted requester checks ✓ Mitigated
ASH-OCLAW-793 — MEDIUM MCP SSE redirects could forward Authorization headers ✓ Mitigated
ASH-OCLAW-794 — HIGH ClickClack agent-mode dispatch could ignore toolsAllow ✓ Mitigated
ASH-OCLAW-795 — MEDIUM MS Teams outbound requests could leak Bot Framework tokens ✓ Mitigated
ASH-OCLAW-796 — MEDIUM Bot Framework serviceUrl validation could leak bot tokens ✓ Mitigated
ASH-OCLAW-797 — MEDIUM QQBot media upload could reach untrusted remote URLs ✓ Mitigated
ASH-OCLAW-798 — HIGH QQBot exec approvals could allow non-allowlisted senders ✓ Mitigated
ASH-OCLAW-799 — HIGH device.pair.approve could bypass role-management checks ✓ Mitigated
ASH-OCLAW-800 — HIGH Hooks allowedAgentIds could be bypassed with blank agent IDs ✓ Mitigated
ASH-OCLAW-801 — MEDIUM WebSocket auth attempts could avoid non-browser rate limits ✓ Mitigated
ASH-OCLAW-802 — MEDIUM ClickClack allowFrom could allow non-allowlisted commands ✓ Mitigated
ASH-OCLAW-803 — HIGH Setup-mode discovery could load untrusted workspace plugins ✓ Mitigated
ASH-OCLAW-804 — HIGH Device-pair approval could expose node system.run early ✓ Mitigated
ASH-OCLAW-805 — MEDIUM [AgentGG] MS Teams allowFrom could bind to mutable display names ✓ Mitigated
ASH-OCLAW-806 — MEDIUM Skill command dispatch could skip effective tool policy ✓ Mitigated
ASH-OCLAW-807 — HIGH Browser act route could miss current-tab URL checks ✓ Mitigated
ASH-OCLAW-808 — HIGH Browser snapshot routes could miss post-navigation SSRF checks ✓ Mitigated
ASH-OCLAW-809 — HIGH Exec allowlist glob matching could allow traversal bypasses ✓ Mitigated
ASH-OCLAW-810 — HIGH Identity-bearing HTTP callers could reach admin-scoped tools ✓ Mitigated
ASH-OCLAW-811 — MEDIUM Remote media URLs could slow-read exhaust tool workers ✓ Mitigated
ASH-OCLAW-812 — MEDIUM Trajectory export could skip broad credential redaction ✓ Mitigated
ASH-OCLAW-813 — HIGH MS Teams safeFetch could race DNS rebinding checks ✓ Mitigated
ASH-OCLAW-814 — HIGH HTTP Canvas responses could forge trusted A2UI actions ✓ Mitigated
ASH-OCLAW-815 — HIGH Node exec approvals could use different gateway and node environments ✓ Mitigated
ASH-OCLAW-816 — HIGH Workspace dotenv files could override provider credentials ✓ Mitigated
Page of 1

14 Hermes Agent CVEs — 13 mitigated, 1 partial

All CVEs disclosed in Hermes Agent (March–July 2026) are documented and tracked by AgentShroud’s defense-in-depth architecture.

CVE CVSS Title Status
ASH-HERMES-014 5.9 MEDIUM Dependency CVE: python-olm (libolm) side-channel/signature weaknesses -- permanently unfixed upstream ⚠ Partial
ASH-HERMES-005 8.1 HIGH OS Command Injection in terminal_tool approval.py detect_dangerous_command ✓ Mitigated
ASH-HERMES-013 7.8 HIGH Dependency CVE: Pillow OS command injection via Image.show() on Windows ✓ Mitigated
ASH-HERMES-004 6.1 HIGH Information Disclosure via _make_run_env in local.py ✓ Mitigated
ASH-HERMES-006 5.6 MEDIUM Missing Authentication on Hermes Agent API Endpoints (GHSA-r7hr-pvjh-r4p3) ✓ Mitigated
ASH-HERMES-007 5.6 MEDIUM Missing Authentication on Hermes Agent Management Endpoints ✓ Mitigated
ASH-HERMES-003 5.3 MEDIUM WebUI Path Traversal via Workspace Path Manipulation ✓ Mitigated
ASH-HERMES-008 5.3 MEDIUM Dependency CVE: pyasn1 unbounded-exponent REAL decode DoS ✓ Mitigated
ASH-HERMES-010 5.3 MEDIUM Dependency CVE: pydantic-settings NestedSecretsSettingsSource symlink escape ✓ Mitigated
ASH-HERMES-012 5.3 MEDIUM Dependency CVE: httplib2 unbounded gzip/deflate decompression DoS ✓ Mitigated
ASH-HERMES-002 4.8 MEDIUM Symlink Following in File Tools _check_sensitive_path ✓ Mitigated
ASH-HERMES-009 4.5 MEDIUM Dependency CVE: PyNaCl bundled libsodium Ed25519 point-validation gap ✓ Mitigated
ASH-HERMES-001 4.0 MEDIUM Path Traversal in WeChat Work Adapter ✓ Mitigated
ASH-HERMES-011 3.3 LOW Dependency CVE: Pygments AdlLexer catastrophic-backtracking ReDoS ✓ Mitigated
Page of 1

AgentShroud Infrastructure

Ships like a product, not a science project. Every release goes through a guarded blue/green deploy with automatic rollback, is validated across every runtime before it lands, and passes only on a genuinely healthy runtime. And because a security product must not hide its own vulnerabilities behind a green check, we publish our container-image CVE posture in full: as of v1.3.0, every container CVE with an available in-place fix is mitigated, verified against a fresh Trivy database (2026-07-15). Nothing is suppressed — the findings that remain have no upstream fix yet, and we enumerate each one with reachability analysis.

🟦

Blue/green canary deploy

A new ref is deployed to a GREEN canary and validated by post-deploy-check.sh before the BLUE production instance is ever touched. A failed health check auto-rolls-back to the previous ref; a pre-deploy-<UTC> tag anchors every rollback. No eval, refuses to run as the prod user.

✓ Live-validated on marvin

🖧

Multi-host validated deploys

One command runs the suite across every host over SSH and aggregates per-host PASS / FAIL / UNREACHABLE — a change is validated on every runtime (x86 + arm) before a coordinated rollout. A flaky host never aborts the sweep.

✓ Shipped

🩺

Honest health gates

Deploys pass only when the gateway serves /status 200 and the Docker network is clean. Startup warnings are curated and documented as known-benign — a green deploy means a genuinely healthy runtime, not a suppressed one.

✓ Enforced

Our own images, reported in full

Alongside the OpenClaw and Hermes application-CVE registries above, we report our own container images — base OS packages and baked-in tool binaries — exactly as Trivy finds them, including items with no upstream fix. Measured against a single fresh Trivy DB snapshot (2026-07-15), v1.3.0 lands every in-place fix that exists: 516 HIGH (down from 535, −19) and 48 CRITICAL across all four images. The CRITICALs do not fall because every one of them is a Debian base-OS package marked fix: NONE upstream — there is no patched build to apply. We show that honestly rather than paint it green.

Image Critical High Status
agentshroud-gateway (v1.3.0) 5 132 ✓ HIGH −16 vs 148 — slsa-verifier rebuilt from source (19H → 3H)
agentshroud-openclaw (v1.3.0) 17 137 ◐ no in-place fix left — residue is Debian fix: NONE + closed-source op
agentshroud-hermes (v1.3.0) 19 181 ◐ HIGH −3 vs 184 (venv patch bumps) — bulk is the upstream vendored base
agentshroud-voice-gateway (v1.3.0) 7 66 ◐ zero fixable — every finding is Debian fix: NONE

Trivy CRITICAL + HIGH, all four images rebuilt & rescanned against one fresh DB snapshot (2026-07-15) so the deltas are apples-to-apples. After every in-place fix, 0 fixable CRITICAL-or-HIGH findings remain — every residual has FixedVersion: (none) at the OS layer, is inside a closed-source binary (op), is a vendored sub-binary only its parent project can rebuild (falcoctl in Falco 0.44.1, latest), or is a core-dependency major/minor jump that would break an upstream-pinned agent. None is suppressed; all remain scanner-visible. Full per-CVE enumeration in docs/security/cve-mitigation-matrix.md.

📦

OS packages

gpg, openssh, git, zlib, libc6, expat, curl and base libraries. Every base image is at its latest pinned digest with apt upgrade in each package-manager stage. The residual are Debian packages marked status: affected, fix: NONE upstream — no patched build exists in the suite to apply.

✓ Mitigated where a fix exists

🔧

Scanner-tool binaries

cosign, docker CLI, fluent-bit, Falco and Wazuh were bumped to patched releases; slsa-verifier was rebuilt from source against patched Go with extended dependency overrides (x/crypto, x/net, go-jose, otel, rekor), taking its HIGH findings 19 → 3. The 3 left are a compile-break pair (fulcio/timestamp-authority) with no in-place fix; falcoctl is a vendored sub-binary only the Falco project can rebuild.

✓ Every available in-place fix applied

🔍

Zero suppression

The health-report scan reports every severity. .trivyignore is zeroed — the sole prior entry (CVE-2021-23358) was retired by the Wazuh 4.14.5 bump, not suppressed; a fresh no-ignorefile scan confirms it is absent. No CVE with an available fix is hidden. Accepting a known-fixable vuln in a security tool would be theater.

✓ Reported honestly

Standards alignment

Built for regulated and OT-adjacent environments from the start.

Self-assessed alignment against publicly available framework guidance — not a certified or independently audited Security Level. Overall system rating: SL 2, with SL 3 capability reached in the specific Foundational Requirements below. See the full compliance matrix for methodology.

IEC 62443
FR3 → SL3
System Integrity — Cosign, Trivy, Falco, Semgrep
IEC 62443
FR6 → SL3
Audit Logging — SHA-256 hash chain + Wazuh SIEM
IEC 62443
FR7 → SL2
Resource Availability — agent-induced exhaustion hardening
OWASP LLM Top 10
Full coverage
LLM01 – LLM10 all addressed across the 7 defense layers
NIST CSF 2.0
All 6 functions
Govern, Identify, Protect, Detect, Respond, Recover — all mapped
CIS Docker
Benchmark 4.8
Setuid/setgid bits stripped, non-root container user
CSA MAESTRO
All 7 layers
L1 model integrity through L7 governance — mapped to AgentShroud's 7 defense layers
OWASP Agentic AI
ASI-01 – ASI-10
Prompt injection, tool abuse, memory poisoning, egress bypass — all addressed
MITRE ATLAS
Full coverage
ML adversarial techniques — PromptGuard, Trivy, Wazuh FIM, audit chain
NIST AI RMF 1.0
All 4 functions
Govern → RBAC/consent; Map → CVE registry; Measure → scorecard; Manage → approval queue
NIST SP 800-190
All 5 risk areas
Image signing, vulnerability scanning, config hardening, network policy, runtime monitoring
DISA Container STIG
CAT I + CAT II
Non-root user, no privileged mode, read-only root FS, resource limits enforced
EU AI Act
Partial coverage
Art 14 human oversight (ApprovalQueue), Art 13 transparency (AuditChain), Art 9 risk management (CVE registry)
ISO/IEC 42001:2023
Partial coverage
§6.1 risk controls, §8 operational controls (ToolACL, ApprovalQueue), §9 monitoring (scorecard)

Up in four commands

Self-hosted Docker stack. No external cloud dependencies.

bash — macOS (Colima) · Linux (Docker CE)
git clone https://github.com/idallasjlabs/agentshroud.git
cd agentshroud

# Store credentials in your OS backend (prompts interactively)
# Auto-detects: 1Password CLI → macOS Keychain → Linux secret-tool → ~/.agentshroud/secrets/
./docker/setup-secrets.sh store

# Bring up the stack — secrets are extracted ephemerally and cleaned up on exit
scripts/asb up

Prerequisites

  • ✓ Git
  • ✓ Docker Engine (Colima on macOS)
  • ✓ Claude OAuth token sk-ant-oat01-...
  • ✓ Telegram bot token
  • ✓ macOS Keychain (built-in, no setup needed)
  • ✓ Python 3 (for key generation)
  • Optional: 1Password CLI (for teams / shared credentials)

asb commands

asb upStart the stack
asb downStop the stack
asb rebuildRebuild images and restart
asb statusShow container health
asb logs [service]Tail logs
asb testRun test suite inside gateway

Full setup guide in the README. Works on macOS (Colima) and Linux (Docker CE).

Open source. MIT licensed.

If your team is deploying AI agents against real infrastructure, the governance gap is real. AgentShroud closes it.

🎙️ Introduction to AgentShroud — podcast episode