pkg:npm/openclaw
共 405 筆 CVECRITICAL21HIGH111MEDIUM135LOW12
✅ 檢查你的版本
所有已知漏洞
- CRITICAL9.9CVE-2026-22172OpenClaw: WebSocket shared-auth connections could self-declare elevated scopesfrom 0, < 2026.3.12
- CRITICAL9.9CVE-2026-28466OpenClaw Vulnerable to Remote Code Execution via Node Invoke Approval Bypass in Gatewayfrom 0, < 2026.2.14
- from 0, < 2026.4.15
- CRITICAL9.8CVE-2026-41386OpenClaw: Unbound bootstrap setup codes allow privilege escalation during pairingfrom 0, < 2026.3.22
- CRITICAL9.8CVE-2026-33578OpenClaw: Google Chat and Zalouser group sender allowlist bypass via policy downgradefrom 0, < 2026.3.28
- CRITICAL9.8CVE-2026-33577OpenClaw: node.pair.approve missing callerScopes validation allows low-privilege operator to approve malicious nodesfrom 0, < 2026.3.28
- from 0, < 2026.3.28
- CRITICAL9.8CVE-2026-28469OpenClaw Google Chat shared-path webhook target ambiguity allowed cross-account policy-context misroutingfrom 0, < 2026.2.14
- CRITICAL9.8CVE-2026-28454OpenClaw has a potential access-group authorization bypass if channel type lookup failsfrom 0, < 2026.2.1
- CRITICAL9.8CVE-2026-28470OpenClaw has an exec allowlist bypass via command substitution/backticks inside double quotesfrom 0, < 2026.2.2
- CRITICAL9.8CVE-2026-28391OpenClaw's Windows cmd.exe parsing may bypass exec allowlist/approval gatingfrom 0, < 2026.2.2
- CRITICAL9.8CVE-2026-28472OpenClaw's gateway connect could skip device identity checks when auth.token was present but not yet validatedfrom 0, < 2026.2.2
- CRITICAL9.6CVE-2026-41397OpenClaw: OpenShell Mirror Sync — Sandbox Escape via Unrestricted File Sync + Symlink Traversalfrom 0, < 2026.3.31
- CRITICAL9.6CVE-2026-41294OpenClaw has a CWD `.env` environment variable injection which bypasses host-env policy and allows config takeoverfrom 0, < 2026.3.28
- CRITICAL9.6CVE-2026-41387OpenClaw's incomplete host env sanitization blocklist allows supply-chain redirection via package-manager env overridesfrom 0, < 2026.3.22
- CRITICAL9.4CVE-2026-32916OpenClaw: Plugin subagent routes could bypass gateway authorization with synthetic admin scopes>= 2026.3.7, < 2026.3.11
- CRITICAL9.4CVE-2026-28446OpenClaw has an inbound allowlist policy bypass in voice-call extension (empty caller ID + suffix matching)from 0, < 2026.2.2
- CRITICAL9.3CVE-2026-32913OpenClaw: fetch-guard forwards custom authorization headers across cross-origin redirectsfrom 0, < 2026.3.7
- CRITICAL9.1CVE-2026-43534OpenClaw: Agent hook events could enqueue trusted system events from unsanitized external inputfrom 0, < 2026.4.10
- >= 2026.4.7, < 2026.4.14
- CRITICAL9.0CVE-2026-32038OpenClaw has a sandbox network isolation bypass via docker.network=container:<id>from 0, < 2026.2.24
- >= 2026.2.23, < 2026.4.12
- >= 2026.4.5, < 2026.4.10
- HIGH8.8CVE-2026-43569OpenClaw: Workspace provider auth choices could auto-enable untrusted provider pluginsfrom 0, < 2026.4.9
- HIGH8.8CVE-2026-43571OpenClaw: Channel setup catalog lookups could include untrusted workspace plugin shadowsfrom 0, < 2026.4.10
- from 0, < 2026.4.9
- HIGH8.8CVE-2026-43584OpenClaw: Exec environment denylist missed high-risk interpreter startup variablesfrom 0, < 2026.4.10
- HIGH8.8CVE-2026-42426OpenClaw `node.pair.approve` placed in `operator.write` scope instead of `operator.pairing` allows unprivileged pairing approvalfrom 0, < 2026.4.8
- HIGH8.8CVE-2026-42422OpenClaw `device.token.rotate` mints tokens for unapproved roles, bypassing device role-upgrade pairingfrom 0, < 2026.4.8
- HIGH8.8CVE-2026-41378OpenClaw: Paired node escalates to gateway RCE via unrestricted node.event agent dispatchfrom 0, < 2026.3.31
- from 0, < 2026.3.31
- HIGH8.8CVE-2026-41303OpenClaw: Discord text `/approve` bypasses `channels.discord.execApprovals.approvers` and allows non-approvers to resolve pending exec approvalsfrom 0, < 2026.3.28
- from 0, < 2026.3.22
- HIGH8.8CVE-2026-35666OpenClaw's system.run allowlist can be bypassed through an unregistered time dispatch wrapperfrom 0, < 2026.3.22
- HIGH8.8CVE-2026-32010In OpenClaw, manually adding sort to tools.exec.safeBins could bypass allowlist approval via --compress-programfrom 0, < 2026.2.22
- HIGH8.8CVE-2026-28363OpenClaw is vulnerable to validation bypass through GNU long-option abbreviations in allowlist modefrom 0, < 2026.2.23
- HIGH8.8CVE-2026-28363OpenClaw is vulnerable to validation bypass through GNU long-option abbreviations in allowlist modefrom 0, < 2026.2.23
- HIGH8.8CVE-2026-28460OpenClaw's system.run allowlist bypass via shell line-continuation command substitutionfrom 0, < 2026.2.22
- from 0, < 2026.2.21
- HIGH8.8CVE-2026-32023OpenClaw's dispatch-wrapper depth-cap mismatch can bypass shell-wrapper approval gating in system.run allowlist modefrom 0, < 2026.2.24
- HIGH8.8CVE-2026-32013OpenClaw gateway agents.files symlink escape allowed out-of-workspace file read/writefrom 0, < 2026.2.25
- HIGH8.8CVE-2026-32060OpenClaw has a path traversal in apply_patch could write/delete files outside the workspacefrom 0, < 2026.2.14
- HIGH8.8CVE-2026-29610OpenClaw: Command hijacking via unsafe PATH handling (bootstrapping + node-host PATH overrides)from 0, < 2026.2.14
- from 0, < 2026.4.22
- from 0, < 2026.3.24
- HIGH8.6CVE-2026-32974OpenClaw: Feishu webhook mode accepted forged events when only `verificationToken` was configuredfrom 0, < 2026.3.12
- HIGH8.6CVE-2026-28451OpenClaw has two SSRF via sendMediaFeishu and markdown image fetching in Feishu extensionfrom 0, < 2026.2.14
- from 0, < 2026.4.10
- HIGH8.4CVE-2026-32918`OpenClaw: session_status` let sandboxed subagents access parent or sibling session statefrom 0, < 2026.3.11
- from 0, < 2026.1.20
- from 0, < 2026.2.14
- HIGH8.2CVE-2026-43526OpenClaw: QQBot reply media URL handling could trigger SSRF and re-upload fetched bytesfrom 0, < 2026.4.12
- from 0, < 2026.3.31
- HIGH8.2CVE-2026-41394OpenClaw: Unauthenticated plugin-auth HTTP routes receive operator runtime scopesfrom 0, < 2026.3.31
- HIGH8.1CVE-2026-43585OpenClaw: Gateway HTTP endpoints re-resolve bearer auth after SecretRef rotationfrom 0, < 2026.4.15
- HIGH8.1CVE-2026-42431OpenClaw `node.invoke(browser.proxy)` bypasses `browser.request` persistent profile-mutation guardfrom 0, < 2026.4.8
- HIGH8.1CVE-2026-41364OpenClaw: SSH sandbox tar upload follows symlinks, enabling arbitrary file write on remote hostfrom 0, < 2026.3.31
- HIGH8.1CVE-2026-35653OpenClaw: `browser.request` still allows `POST /reset-profile` through the `operator.write` surfacefrom 0, < 2026.3.24
- HIGH8.1CVE-2026-35645OpenClaw: Gateway Plugin Subagent Fallback `deleteSession` Uses Synthetic `operator.admin`from 0, < 2026.3.28
- HIGH8.1CVE-2026-32302OpenClaw: Untrusted web origins can obtain authenticated operator.admin access in trusted-proxy modefrom 0, < 2026.3.11
- from 0, < 2026.2.14
- >= 2026.1.20, < 2026.2.1
- HIGH8.0CVE-2026-32978OpenClaw: Unrecognized script runners could bypass `system.run` approval integrityfrom 0, < 2026.3.11
- HIGH8.0CVE-2026-32014OpenClaw: Node reconnect metadata spoofing could bypass platform-based node command policyfrom 0, < 2026.2.26
- HIGH7.8CVE-2026-45004OpenClaw vulnerable to arbitrary code execution via attacker-controlled setup-api.js loaded from cwd during env-key resolutionfrom 0, < 2026.4.23
- HIGH7.8CVE-2026-44118OpenClaw: MCP loopback owner context is derived from server-issued bearer tokensfrom 0, < 2026.4.22
- HIGH7.8CVE-2026-44114OpenClaw: Workspace dotenv could override runtime-control environment variablesfrom 0, < 2026.4.20
- HIGH7.8CVE-2026-41384OpenClaw Has Incomplete Fix for CVE-2026-4039: CLI Backend Environment Variable Injection via Workspace Configfrom 0, < 2026.3.24
- from 0, < 2026.3.31
- HIGH7.8CVE-2026-41336OpenClaw: Workspace `.env` can override the bundled hooks root and load attacker hook codefrom 0, < 2026.3.31
- HIGH7.8CVE-2026-32015OpenClaw's `tools.exec.safeBins` PATH-hijack allowed trojan binaries to bypass allowlist checks>= 2026.1.21, < 2026.2.19
- HIGH7.8CVE-2026-32032OpenClaw's shell env fallback trusts unvalidated SHELL path from host environmentfrom 0, < 2026.2.22
- HIGH7.8CVE-2026-32016OpenClaw: macOS optional allowlist basename matching could bypass path-based policyfrom 0, < 2026.2.22
- HIGH7.7CVE-2026-43576OpenClaw: CDP /json/version WebSocket URL could pivot to untrusted second-hop targetsfrom 0, < 2026.4.5
- HIGH7.7CVE-2026-43580OpenClaw: Browser press/type interaction routes missed complete navigation guard coveragefrom 0, < 2026.4.10
- HIGH7.7CVE-2026-43573OpenClaw: Existing-session browser interaction routes bypassed SSRF policy enforcementfrom 0, < 2026.4.10
- from 0, < 2026.4.14
- >= 2026.4.7, < 2026.4.10
- HIGH7.7CVE-2026-42436OpenClaw: Browser snapshot and screenshot routes could expose internal page content after navigationfrom 0, < 2026.4.14
- HIGH7.7CVE-2026-35668OpenClaw has Sandbox Media Root Bypass via Unnormalized `mediaUrl` / `fileUrl` Parameter Keys (CWE-22)from 0, < 2026.3.24
- from 0, < 2026.2.21
- HIGH7.6CVE-2026-41297OpenClaw: Marketplace Plugin Download Follows Redirects Without SSRF Protectionfrom 0, < 2026.3.31
- from 0, < 2026.2.26
- from 0, < 2026.2.14
- from 0, < 2026.2.14
- HIGH7.5CVE-2026-42423OpenClaw: strictInlineEval explicit-approval boundary bypassed by approval-timeout fallback on gateway and node exec hostsfrom 0, < 2026.4.8
- HIGH7.5CVE-2026-41405OpenClaw: MS Teams webhook parses body before JWT validation, enabling unauthenticated resource exhaustionfrom 0, < 2026.3.31
- HIGH7.5CVE-2026-34503OpenClaw's device removal and token revocation do not terminate active WebSocket sessionsfrom 0, < 2026.3.28
- from 0, < 2026.3.22
- HIGH7.5CVE-2026-32980OpenClaw Telegram webhook request bodies were read before secret validation, enabling unauthenticated resource exhaustionfrom 0, < 2026.3.13
- HIGH7.5CVE-2026-32025OpenClaw's browser-origin WebSocket auth hardening gap could enable loopback password brute-force chainsfrom 0, < 2026.2.25
- HIGH7.5CVE-2026-32056OpenClaw's shell startup env injection bypasses system.run allowlist intent (RCE class)from 0, < 2026.2.22
- HIGH7.5CVE-2026-32033OpenClaw has a workspace-only sandbox guard mismatch for @-prefixed absolute pathsfrom 0, < 2026.2.24
- HIGH7.5CVE-2026-32034OpenClaw has an opt-in insecure Control UI auth over plaintext HTTP could allow privileged accessfrom 0, < 2026.2.21
- HIGH7.5CVE-2026-32011OpenClaw has pre-auth webhook body parsing that can enable unauthenticated slow-request DoSfrom 0, < 2026.3.2
- from 0, < 2026.2.19
- HIGH7.5CVE-2026-32062OpenClaw voice-call media stream validated streams after upgrade, which could allow pre-start unauthenticated sockets to increase resource pressurefrom 0, < 2026.2.22
- from 0, < 2026.2.14
- HIGH7.5CVE-2026-28462OpenClaw has a path traversal in browser trace/download output paths may allow arbitrary file writesfrom 0, < 2026.2.13
- HIGH7.5CVE-2026-28478OpenClaw affected by denial of service via unbounded webhook request body bufferingfrom 0, < 2026.2.13
- from 0, < 2026.2.14
- HIGH7.5CVE-2026-26324OpenClaw has a SSRF guard bypass via full-form IPv4-mapped IPv6 (loopback / metadata reachable)from 0, < 2026.2.14
- from 0, < 2026.2.14
- HIGH7.5CVE-2026-26319OpenClaw is Missing Webhook Authentication in Telnyx Provider Allows Unauthenticated Requestsfrom 0, < 2026.2.14
- from 0, < 2026.2.13
- HIGH7.5CVE-2026-25474OpenClaw has a Telegram webhook request forgery (missing `channels.telegram.webhookSecret`) → auth bypassfrom 0, < 2026.2.1
- HIGH7.5CVE-2026-28458OpenClaw's Browser Relay /cdp websocket is missing auth which could allow cross-tab cookie access>= 2026.1.20, < 2026.2.1
- HIGH7.4CVE-2026-35629OpenClaw: SSRF via Unguarded Configured Base URLs in Multiple Channel Extensions (Incomplete Fix for CVE-2026-28476)from 0, < 2026.3.28
- HIGH7.4CVE-2026-28481OpenClaw MS Teams inbound attachment downloader leaks bearer tokens to allowlisted suffix domainsfrom 0, < 2026.2.1
- from 0, < 2026.3.31
- HIGH7.3CVE-2026-41355OpenClaw: OpenShell `mirror` mode can convert untrusted sandbox files into explicitly enabled workspace hooks and execute them on the host during gateway startupfrom 0, < 2026.3.28
- from 0, < 2026.3.28
- HIGH7.3CVE-2026-41390OpenClaw has a gateway exec allowlist allow-always bypass via unregistered /usr/bin/script wrapperfrom 0, < 2026.3.28
- HIGH7.3CVE-2026-41342OpenClaw: CLI Remote Onboarding Persists Unauthenticated Discovery Endpoint and Exfiltrates Gateway Credentialsfrom 0, < 2026.3.28
- HIGH7.3CVE-2026-32979OpenClaw: Unbound interpreter and runtime commands could bypass node-host approval integrityfrom 0, < 2026.3.11
- HIGH7.3CVE-2026-28448OpenClaw Twitch allowFrom is not enforced in optional plugin, unauthorized chat users can trigger agent pipeline>= 2026.1.29, < 2026.2.1
- HIGH7.2CVE-2026-28456OpenClaw affected by potential code execution via unsafe hook module path handling in Gateway>= 2026.1.5, < 2026.2.14
- HIGH7.2CVE-2026-26325OpenClaw Node host system.run rawCommand/command mismatch can bypass allowlist/approvalsfrom 0, < 2026.2.14
- HIGH7.2CVE-2026-28473OpenClaw authorization bypass: operator.write can resolve exec approvals via chat.send -> /approvefrom 0, < 2026.2.2
- HIGH7.1CVE-2026-42428OpenClaw B-M3: ClawHub package downloads are not enforced with integrity verificationfrom 0, < 2026.4.8
- HIGH7.1CVE-2026-41359OpenClaw: Gateway operator.write Can Reach Admin-Class Telegram Config and Cron Persistence via sendfrom 0, < 2026.3.28
- HIGH7.1CVE-2026-41347OpenClaw: HTTP operator endpoints lack browser-origin validation in trusted-proxy modefrom 0, < 2026.3.31
- HIGH7.1CVE-2026-41299OpenClaw: Gateway chat.send ACP-only provenance guard could be bypassed by client identity spoofingfrom 0, < 2026.3.28
- HIGH7.1CVE-2026-35621OpenClaw: Gateway operator.write Can Reach Admin-Class Channel Allowlist Persistence via chat.sendfrom 0, < 2026.3.24
- HIGH7.1CVE-2026-35632OpenClaw: Symlink Traversal via IDENTITY.md appendFile in agents.create/update (Incomplete Fix for CVE-2026-32013)from 0, <= 2026.2.22
- HIGH7.1CVE-2026-32971OpenClaw: Node-host approvals could show misleading shell payloads instead of the executed argvfrom 0, < 2026.3.11
- HIGH7.1CVE-2026-32057OpenClaw has a Trusted-proxy Control UI pairing bypass which allows unpaired node sessionsfrom 0, < 2026.2.25
- HIGH7.1CVE-2026-32017OpenClaw exec allowlist safeBins short-option bypass could permit arbitrary file writefrom 0, < 2026.2.19
- HIGH7.1CVE-2026-28457OpenClaw's sandbox skill mirroring path traversal vulnerability could write outside the sandbox workspacefrom 0, < 2026.2.14
- >= 2026.1.29-beta.1, < 2026.2.14
- HIGH7.1CVE-2026-26317OpenClaw affected by cross-site request forgery (CSRF) through loopback browser mutation endpointsfrom 0, < 2026.2.14
- from 0, < 2026.2.12
- MEDIUM6.9CVE-2026-35654OpenClaw: MS Teams Feedback Invocation Bypasses Sender Allowlists and Records Unauthorized Session Feedbackfrom 0, < 2026.3.28
- MEDIUM6.9CVE-2026-35664OpenClaw: Feishu Raw Card Send Surface Can Mint Legacy Card Callbacks That Bypass DM Pairingfrom 0, < 2026.3.28
- MEDIUM6.9CVE-2026-32041OpenClaw: Browser control startup could continue unauthenticated after auth bootstrap failurefrom 0, < 2026.3.1
- MEDIUM6.8CVE-2026-43535OpenClaw: Collect-mode queue batches could reuse the last sender authorization contextfrom 0, < 2026.4.14
- MEDIUM6.8CVE-2026-32005OpenClaw: Slack interactive callbacks could skip configured sender checks in some shared-workspace flowsfrom 0, < 2026.2.25
- MEDIUM6.8CVE-2026-32007OpenClaw: Experimental apply_patch may bypass workspace-only checks in opt-in sandbox mounts (off by default)from 0, < 2026.2.23
- MEDIUM6.7CVE-2026-29608OpenClaw's Node system.run approval hardening wrapper semantic drift can execute unintended local scripts>= 2026.3.1, < 2026.3.2
- >= 2026.1.12, < 2026.2.13
- MEDIUM6.6CVE-2026-32003OpenClaw has system.run shell-wrapper env injection via SHELLOPTS/PS4 can bypass allowlist intent (RCE)from 0, < 2026.2.22
- >= 2026.3.22, < 2026.4.5
- MEDIUM6.5CVE-2026-42433OpenClaw: Matrix profile config persistence was reachable from operator.write message toolsfrom 0, < 2026.4.10
- from 0, < 2026.4.12
- MEDIUM6.5CVE-2026-42430OpenClaw: Strict browser SSRF bypass in Playwright redirect handling leaves private targets reachablefrom 0, < 2026.4.8
- MEDIUM6.5CVE-2026-41408OpenClaw: Tlon media downloads can bypass core safety limits and exhaust diskfrom 0, < 2026.3.31
- MEDIUM6.5CVE-2026-41369OpenClaw: Host exec environment sanitization misses package, registry, Docker, compiler, and TLS override variablesfrom 0, < 2026.3.31
- from 0, < 2026.3.31
- MEDIUM6.5CVE-2026-41385OpenClaw Nostr privateKey config redaction bypass leaks plaintext signing key via config.getfrom 0, < 2026.3.31
- MEDIUM6.5CVE-2026-33580OpenClaw's Nextcloud Talk webhook missing rate limiting on shared secret authenticationfrom 0, < 2026.3.28
- MEDIUM6.5CVE-2026-34508Duplicate Advisory: OpenClaw has Bypass in Webhook Rate Limiting via Pre-Authentication Secret Validationfrom 0, < 2026.3.12
- MEDIUM6.5CVE-2026-35658OpenClaw: Image Tool `tools.fs.workspaceOnly` Bypass via Sandbox Bridge Mountsfrom 0, < 2026.3.2
- MEDIUM6.5CVE-2026-35656OpenClaw: Forwarding header spoofing bypasses gateway.trustedProxies origin detectionfrom 0, < 2026.3.22
- MEDIUM6.5CVE-2026-35652OpenClaw: Mattermost callback dispatch allowed non-allowlisted sender actionsfrom 0, < 2026.3.22
- MEDIUM6.5CVE-2026-32021OpenClaw has a Feishu allowFrom authorization bypass via display-name collisionfrom 0, < 2026.2.22
- MEDIUM6.5CVE-2026-32008OpenClaw browser navigation guard allowed non-network URL schemes, enabling authenticated browser-tool users to access file:// local filesfrom 0, < 2026.2.21
- MEDIUM6.5CVE-2026-32004OpenClaw has encoded-path auth bypass in plugin `/api/channels` route classificationfrom 0, < 2026.3.2
- MEDIUM6.5CVE-2026-32027OpenClaw DM pairing-store identities could satisfy group allowlist authorizationfrom 0, < 2026.2.26
- MEDIUM6.5CVE-2026-32036OpenClaw has gateway plugin auth bypass via encoded dot-segment traversal in protected /api/channels pathsfrom 0, < 2026.2.26
- from 0, < 2026.2.24
- from 0, < 2026.2.15
- MEDIUM6.5CVE-2026-29606OpenClaw Twilio voice-call webhook auth bypass when ngrok loopback compatibility is enabledfrom 0, < 2026.2.14
- MEDIUM6.5CVE-2026-28452OpenClaw affected by denial of service through unguarded archive extraction allowing high expansion/resource abuse (ZIP/TAR)from 0, < 2026.2.14
- MEDIUM6.5CVE-2026-26328OpenClaw iMessage group allowlist authorization inherited DM pairing-store identitiesfrom 0, < 2026.2.14
- MEDIUM6.5CVE-2026-28471OpenClaw has a Matrix allowlist bypass via displayName and cross-homeserver localpart matching>= 2026.1.14-1, < 2026.2.2
- MEDIUM6.5CVE-2026-28395OpenClaw's Chrome extension relay binds publicly due to wildcard treated as loopback>= 2026.1.14-1, < 2026.2.12
- from 0, < 2026.1.30
- MEDIUM6.4CVE-2026-22169OpenClaw's non-default safeBins sort configuration can bypass intended allowlist approval constraintsfrom 0, < 2026.2.22
- MEDIUM6.4CVE-2026-29607OpenClaw's allow-always wrapper persistence could bypass future approvals and enable command executionfrom 0, < 2026.2.22
- MEDIUM6.3CVE-2026-43582OpenClaw: Browser SSRF hostname validation could be bypassed by DNS rebindingfrom 0, < 2026.4.10
- MEDIUM6.3CVE-2026-41302OpenClaw: SSRF via Unguarded `fetch()` in Marketplace Plugin Download and Ollama Model Discoveryfrom 0, < 2026.3.31
- from 0, < 2026.3.11
- MEDIUM6.3CVE-2026-32921OpenClaw's system.run approvals did not bind mutable script operands across approval and executionfrom 0, < 2026.3.8
- MEDIUM6.2CVE-2026-33581OpenClaw's message tool media parameter bypasses tool policy filesystem isolationfrom 0, < 2026.3.24
- MEDIUM6.2CVE-2026-28450OpenClaw's unauthenticated Nostr profile HTTP endpoints allow remote profile/config tamperingfrom 0, < 2026.2.12
- MEDIUM6.1CVE-2026-41373OpenClaw: Incomplete host-env-security-policy allows untrusted model to substitute compiler binaries via env overridesfrom 0, < 2026.3.31
- MEDIUM6.1CVE-2026-35667OpenClaw has incomplete Fix for CVE-2026-27486: Unvalidated SIGKILL in `!stop` Chat Command via `shell-utils.ts`from 0, < 2026.3.24
- from 0, < 2026.3.7
- MEDIUM6.1CVE-2026-28486OpenClaw vulnerable to path traversal (Zip Slip) in archive extraction during explicit installation commands>= 2026.1.16-2, < 2026.2.14
- MEDIUM6.0CVE-2026-45005OpenClaw's Webhooks SecretRef route secret remains valid after rotation/reloadfrom 0, < 2026.4.23
- MEDIUM6.0CVE-2026-32037OpenClaw's MSTeams attachment redirect handling could bypass configured media host allowlistsfrom 0, < 2026.2.22
- MEDIUM6.0CVE-2026-28393OpenClaw's hook transform module path allows traversal and arbitrary JavaScript module loading>= 2.0.0-beta3, < 2026.2.14
- MEDIUM5.9CVE-2026-35622OpenClaw: Google Chat app-url webhook auth accepted non-deployment add-on principalsfrom 0, < 2026.3.22
- MEDIUM5.9CVE-2026-35670OpenClaw: Synology Chat reply delivery could be rebound through username-based user resolution.from 0, < 2026.3.22
- MEDIUM5.9CVE-2026-32039OpenClaw's typed sender-key matching for toolsBySender prevents identity-collision policy bypassfrom 0, < 2026.2.22
- MEDIUM5.9CVE-2026-32035OpenClaw: Discord voice transcript owner-flag omission could expose owner-only tools in mixed-trust channelsfrom 0, < 2026.3.2
- from 0, < 2026.2.12
- MEDIUM5.9CVE-2026-28477OpenClaw Chutes manual OAuth state validation bypass can cause credential substitutionfrom 0, < 2026.2.14
- from 0, < 2026.2.14
- from 0, < 2026.2.2
- MEDIUM5.9CVE-2026-29613OpenClaw has a webhook auth bypass when gateway is behind a reverse proxy (loopback remoteAddress trust)from 0, < 2026.2.12
- from 0, < 2026.4.20
- MEDIUM5.8CVE-2026-41389OpenClaw: Webchat media embedding enforces local-root containment for tool-result files>= 2026.4.7, < 2026.4.15
- MEDIUM5.8CVE-2026-41372OpenClaw: Trailing-dot localhost CDP hosts could bypass remote loopback protectionsfrom 0, < 2026.4.2
- MEDIUM5.8CVE-2026-27009OpenClaw affected by Stored XSS in Control UI via unsanitized assistant name/avatar in inline script injectionfrom 0, < 2026.2.15
- MEDIUM5.7CVE-2026-40045OpenClaw: Android accepted cleartext remote gateway endpoints and sent stored credentials over ws://from 0, < 2026.4.2
- from 0, < 2026.3.22
- from 0, < 2026.2.22
- MEDIUM5.7CVE-2026-32009OpenClaw: safeBins static default trusted dirs allow writable-dir binary hijack (`jq`)from 0, < 2026.2.24
- MEDIUM5.7CVE-2026-28463OpenClaw exec approvals: safeBins could bypass stdin-only constraints via shell expansionfrom 0, < 2026.2.14
- from 0, < 2026.1.29
- MEDIUM5.5CVE-2026-32044OpenClaw skills-install-download: tar.bz2 extraction bypassed archive safety parity checks (local DoS)from 0, < 2026.3.2
- from 0, < 2026.2.22
- MEDIUM5.5CVE-2026-28482OpenClaw's unsanitized session ID enables path traversal in transcript file operationsfrom 0, < 2026.2.12
- MEDIUM5.5CVE-2026-29612OpenClaw: denial of service through large base64 media files allocating large buffers before limit checksfrom 0, < 2026.2.14
- MEDIUM5.4CVE-2026-41358OpenClaw: Slack thread context could include messages from non-allowlisted sendersfrom 0, < 2026.4.2
- MEDIUM5.4CVE-2026-41298OpenClaw: Read-scoped identity-bearing HTTP clients could kill sessions via /sessions/:sessionKey/killfrom 0, < 2026.4.2
- MEDIUM5.4CVE-2026-41341OpenClaw: Discord Component Interaction Misclassifies Group DM as Direct Messagefrom 0, < 2026.3.31
- from 0, < 2026.3.31
- MEDIUM5.4CVE-2026-41356OpenClaw: Gateway `device.token.rotate` does not terminate active WebSocket sessions after credential rotationfrom 0, < 2026.3.31
- from 0, < 2026.3.31
- MEDIUM5.4CVE-2026-41344OpenClaw: Gateway `operator.write` can reach admin-only persisted `verboseLevel` via `chat.send` `/verbose`from 0, < 2026.3.28
- MEDIUM5.4CVE-2026-35620OpenClaw: Non-owner command-authorized sender can change the owner-only `/send` session delivery policyfrom 0, < 2026.3.24
- MEDIUM5.4CVE-2026-32895OpenClaw: Slack system events bypass sender authorization in member and message subtype handlersfrom 0, < 2026.2.26
- MEDIUM5.4CVE-2026-32001OpenClaw's Node role device-identity bypass allows unauthorized node.event injectionfrom 0, < 2026.2.22
- MEDIUM5.4CVE-2026-32898OpenClaw ACP client has permission auto-approval bypass via untrusted tool metadatafrom 0, < 2026.2.23
- from 0, < 2026.2.15
- MEDIUM5.3CVE-2026-44113OpenClaw: OpenShell FS bridge reads pin and verify the opened file before returning bytesfrom 0, < 2026.4.22
- from 0, < 2026.4.22
- MEDIUM5.3CVE-2026-41915OpenClaw: GIT_DIR and related git plumbing env vars missing from exec env denylist (GHSA-m866-6qv5-p2fg variant)from 0, < 2026.4.8
- MEDIUM5.3CVE-2026-41346OpenClaw: Pairing pending-request caps were enforced per channel instead of per account>= 2026.2.26, < 2026.3.31
- MEDIUM5.3CVE-2026-41301OpenClaw: Forged Nostr DMs could create pairing state before signature verification>= 2026.3.22, < 2026.3.31
- MEDIUM5.3CVE-2026-34425OpenClaw's complex interpreter pipelines could skip exec script preflight validationfrom 0, < 2026.4.2
- MEDIUM5.3CVE-2026-41331OpenClaw: Telegram audio preflight transcription enables resource consumption by unauthorized sendersfrom 0, < 2026.3.31
- MEDIUM5.3CVE-2026-41400OpenClaw: Voice-call still parses large WebSocket frames before start validation (Incomplete fix for CVE-2026-32062)from 0, < 2026.3.31
- MEDIUM5.3CVE-2026-41351OpenClaw: Telnyx Webhook Replay Detection Bypass via Base64 Signature Re-encodingfrom 0, < 2026.3.31
- MEDIUM5.3CVE-2026-41374OpenClaw runs Discord audio preflight transcription before member authorizationfrom 0, < 2026.3.31
- MEDIUM5.3CVE-2026-41343OpenClaw: LINE webhook handler lacks shared pre-auth concurrency budget before signature verificationfrom 0, < 2026.3.31
- MEDIUM5.3CVE-2026-41391OpenClaw: PIP_INDEX_URL and UV_INDEX_URL bypass host exec env sanitization and redirect Python package-index trafficfrom 0, < 2026.3.31
- MEDIUM5.3CVE-2026-41337OpenClaw: Voice-call Plivo replay mutates in-process callback origin before replay rejectionfrom 0, < 2026.3.31
- MEDIUM5.3CVE-2026-41363OpenClaw: Feishu extension resolveUploadInput bypasses file-system sandbox and allows arbitrary file reads via upload_image>= 2026.2.6, < 2026.3.28
- MEDIUM5.3CVE-2026-35665OpenClaw has incomplete Fix for CVE-2026-32011: Feishu Webhook Pre-Auth Body Parsing DoS (Slow-Body / Slowloris Variant)from 0, < 2026.3.24
- MEDIUM5.3CVE-2026-35661OpenClaw: Telegram DM-Scoped Inline Button Callbacks Bypass DM Pairing and Mutate Session Statefrom 0, < 2026.3.28
- from 0, < 2026.3.7
- MEDIUM5.3CVE-2026-32002OpenClaw's image tool bypasses tools.fs.workspaceOnly on sandbox mount paths and exfiltrates out-of-workspace imagesfrom 0, < 2026.2.23
- from 0, < 2026.2.22
- MEDIUM5.3CVE-2026-32029OpenClaw improperly parses X-Forwarded-For behind trusted proxies allows client IP spoofing in security decisionsfrom 0, < 2026.2.21
- MEDIUM5.3CVE-2026-22180OpenClaw: Unified root-bound write hardening for browser output and related path-boundary flowsfrom 0, < 2026.3.2
- MEDIUM5.0CVE-2026-42424OpenClaw: Shared reply MEDIA - paths are treated as trusted and can trigger cross-channel local file exfiltrationfrom 0, < 2026.4.8
- MEDIUM4.9CVE-2026-41332OpenClaw host-env blocklist missing `GIT_TEMPLATE_DIR` and `AWS_CONFIG_FILE` allows code execution via env overridefrom 0, < 2026.3.28
- MEDIUM4.8CVE-2026-35628OpenClaw: Telegram Webhook Missing Guess Rate Limiting Enables Brute-Force Guessing of Weak Webhook Secretfrom 0, <= 2026.3.24
- MEDIUM4.8CVE-2026-35623OpenClaw: BlueBubbles Webhook Missing Rate Limiting Enables Brute-Force Password Guessingfrom 0, <= 2026.3.24
- MEDIUM4.8CVE-2026-32031OpenClaw: /api/channels gateway-auth boundary bypass via path canonicalization mismatchfrom 0, < 2026.2.26
- MEDIUM4.8CVE-2026-32018OpenClaw's serialize sandbox registry writes to prevent races and delete-rollback corruptionfrom 0, < 2026.2.19
- MEDIUM4.8CVE-2026-32896OpenClaw: BlueBubbles beta plugin webhook auth hardening (remove passwordless fallback)from 0, < 2026.2.21
- MEDIUM4.8CVE-2026-32065OpenClaw: system.run approval identity mismatch could execute a different binary than displayedfrom 0, < 2026.2.25
- from 0, < 2026.2.13
- MEDIUM4.8CVE-2026-28392OpenClaw Slack: dmPolicy=open allowed any DM sender to run privileged slash commandsfrom 0, < 2026.2.14
- MEDIUM4.6CVE-2026-41377OpenClaw: Security Scan Failure Does Not Block Plugin Installation (Fail-Open)from 0, < 2026.3.31
- MEDIUM4.6CVE-2026-35659OpenClaw: Bonjour/DNS-SD TXT metadata steers CLI routing after failed service resolutionfrom 0, < 2026.3.22
- MEDIUM4.6CVE-2026-32040OpenClaw Vulnerable to HTML injection via unvalidated image MIME type in data-URL interpolationfrom 0, < 2026.2.23
- MEDIUM4.4CVE-2026-41330OpenClaw: Host exec environment overrides miss proxy, TLS, Docker, and Git TLS controlsfrom 0, < 2026.3.31
- from 0, < 2026.2.17
- MEDIUM4.3CVE-2026-41908OpenClaw: Assistant media route missed scope enforcement for trusted-proxy authorizationfrom 0, < 2026.4.20
- from 0, < 2026.4.8
- MEDIUM4.3CVE-2026-41910OpenClaw: /allowlist omits owner-only enforcement for cross-channel allowlist writesfrom 0, < 2026.4.8
- MEDIUM4.3CVE-2026-41339OpenClaw: Gateway hello snapshots exposed host config and state paths to non-admin clientsfrom 0, < 2026.4.2
- from 0, < 2026.3.24
- >= 2026.2.13, < 2026.3.28
- MEDIUM4.3CVE-2026-35662OpenClaw leaf subagents can bypass controlScope restrictions to send messages to child sessionsfrom 0, < 2026.3.22
- MEDIUM4.3CVE-2026-32006OpenClaw has a BlueBubbles group allowlist mismatch via DM pairing-store fallbackfrom 0, < 2026.2.26
- MEDIUM4.3CVE-2026-32899OpenClaw's Slack reaction/pin sender-policy consistency issue in non-message ingressfrom 0, < 2026.2.25
- from 0, < 2026.2.19
- from 0, < 2026.3.31
- MEDIUM4.2CVE-2026-35617OpenClaw: Google Chat Authz Bypass via Group Policy Rebinding with Mutable Space displayNamefrom 0, < 2026.3.28
- MEDIUM4.2CVE-2026-35624OpenClaw: Nextcloud Talk room allowlist matched colliding room names instead of stable room tokensfrom 0, < 2026.3.22
- MEDIUM4.0CVE-2026-41403OpenClaw: diffs viewer misclassifies proxied remote requests as loopback when `allowRemoteViewer` is disabledfrom 0, < 2026.3.31
- LOW3.7CVE-2026-41913OpenClaw: Concurrent async auth attempts can bypass the intended shared-secret rate-limit budget on Tailscale-capable pathsfrom 0, < 2026.4.4
- LOW3.7CVE-2026-41407OpenClaw: Shared-secret comparison call sites leaked length information through timingfrom 0, < 2026.4.2
- from 0, < 2026.3.31
- from 0, < 2026.3.22
- LOW3.7CVE-2026-32067OpenClaw has cross-account DM pairing authorization bypass via unscoped pairing store accessfrom 0, < 2026.2.26
- LOW3.7CVE-2026-32028OpenClaw: Discord DM reaction ingress missed dmPolicy/allowFrom checks in restricted setupsfrom 0, < 2026.2.25
- LOW3.7CVE-2026-31991OpenClaw has Signal group allowlist authorization bypass via DM pairing-store leakagefrom 0, < 2026.2.26
- LOW3.7CVE-2026-24764OpenClaw Affected by Remote Code Execution via System Prompt Injection in Slack Channel Descriptionsfrom 0, < 2026.2.3
- from 0, < 2026.2.19
- LOW3.3CVE-2026-32020OpenClaw's Control UI Static File Handler Follows Symlinks and Allows Out-of-Root File Readfrom 0, < 2026.2.22
- LOW2.6CVE-2026-32058OpenClaw Node system.run approval context-binding weakness in approval-enabled host=node flowsfrom 0, < 2026.2.26
- LOW2.5CVE-2026-32970OpenClaw: Unavailable local auth SecretRefs could fall through to remote credentials in local modefrom 0, < 2026.3.11
- from 0, < 2026.4.22
- from 0, < 2026.4.22
- —CVE-2026-44991OpenClaw: Owner-enforced commands could accept wildcard channel senders as command ownersfrom 0, < 2026.4.21
- —CVE-2026-44992OpenClaw: Workspace dotenv MiniMax host override could redirect credentialed requests>= 2026.4.5, < 2026.4.20
- —CVE-2026-44995OpenClaw: MCP stdio server env could load dangerous startup variables from workspace configfrom 0, < 2026.4.20
- from 0, < 2026.4.20
- from 0, < 2026.4.20
- —CVE-2026-42438OpenClaw: Sender policy bypass in host media attachment reads allows unauthorized local file disclosure>= 2026.4.9, < 2026.4.10
- from 0, < 2026.4.10
- from 0, < 2026.4.10
- >= 2026.2.22, < 2026.4.12
- —CVE-2026-43568OpenClaw: Memory dreaming config persistence was reachable from operator.write commands>= 2026.4.5, < 2026.4.10
- >= 2026.4.10, < 2026.4.14
- —CVE-2026-43583OpenClaw: Delivery queue recovery could lose group tool-policy context for media replay>= 2026.4.10, < 2026.4.14
- >= 2026.4.9, < 2026.4.10
- from 0, < 2026.4.14
- from 0, < 2026.4.10
- —CVE-2026-40037OpenClaw: `fetchWithSsrFGuard` replays unsafe request bodies across cross-origin redirectsfrom 0, < 2026.4.8
- —CVE-2026-42429OpenClaw: Gateway plugin HTTP `auth: gateway` widens identity-bearing `operator.read` requests into runtime `operator.write`from 0, < 2026.4.8
- from 0, < 2026.4.8
- —CVE-2026-41911OpenClaw: Feishu docx upload_file/upload_image Bypasses Workspace-Only Filesystem Policy (GHSA-qf48-qfv4-jjm9 Incomplete Fix)from 0, < 2026.4.8
- from 0, < 2026.4.8
- from 0, < 2026.4.8
- —CVE-2026-42432OpenClaw: Node Pairing Reconnect Command Escalation Bypasses operator.admin Scope Requirementfrom 0, < 2026.4.8
- from 0, < 2026.4.8
- —CVE-2026-42427OpenClaw: HGRCPATH, CARGO_BUILD_RUSTC_WRAPPER, RUSTC_WRAPPER, and MAKEFLAGS missing from exec env denylist — RCE via build tool env injection (GHSA-cm8v-2vh9-cxf3 class)from 0, < 2026.4.8
- from 0, < 2026.4.2
- —CVE-2026-41295OpenClaw: Untrusted workspace channel shadows could execute during built-in channel setupfrom 0, < 2026.4.2
- —CVE-2026-41398OpenClaw: iOS A2UI bridge trusted generic local-network pages for agent.request dispatchfrom 0, < 2026.4.2
- —CVE-2026-41383OpenClaw: OpenShell mirror mode could delete arbitrary remote directories when roots were mis-scopedfrom 0, < 2026.4.2
- —CVE-2026-41379OpenClaw: Gateway operator.write Can Reach Admin-Class Talk Voice Config Persistence via chat.sendfrom 0, < 2026.3.28
- —CVE-2026-41375OpenClaw: `/phone arm`/`/phone disarm` Bypasses `operator.admin` Scope Check for External Channelsfrom 0, < 2026.3.28
- from 0, < 2026.4.2
- —CVE-2026-41382OpenClaw: Discord voice ingress authorization can be bypassed via channel, name, and stale-role validation gapsfrom 0, < 2026.3.31
- from 0, < 2026.3.31
- from 0, < 2026.3.31
- from 0, < 2026.3.31
- from 0, < 2026.3.31
- —CVE-2026-41404OpenClaw: Incomplete scope-clearing fix allows operator.admin escalation via trusted-proxy auth modefrom 0, < 2026.3.31
- from 0, < 2026.3.31
- from 0, < 2026.3.31
- —CVE-2026-41329OpenClaw: Heartbeat context inheritance bypasses sandbox via senderIsOwner escalationfrom 0, < 2026.3.31
- from 0, < 2026.3.28
- from 0, < 2026.3.28
- —CVE-2026-33579OpenClaw: /pair approve command path omitted caller scope subsetting and reopened device pairing escalationfrom 0, < 2026.3.28
- —CVE-2026-41395OpenClaw: Voice-call Plivo V3 webhook replay key uses unsorted URL, allowing replay via query-parameter reorderingfrom 0, < 2026.3.28
- —CVE-2026-35646OpenClaw: Synology Chat Webhook Pre-Auth Rate-Limit Bypass Enables Brute-Force Guessing of Webhook Tokenfrom 0, < 2026.3.28
- —CVE-2026-35640OpenClaw: Feishu webhook reads and parses unauthenticated request bodies before signature validationfrom 0, < 2026.3.28
- from 0, < 2026.3.25
- —CVE-2026-35647OpenClaw: Matrix Verification Notices Bypass Matrix DM Policy and Reply to Unpaired DM Peersfrom 0, <= 2026.3.24
- —CVE-2026-35669OpenClaw: Gateway Plugin HTTP Auth Grants Unrestricted operator.admin Runtime Scope to All Callersfrom 0, <= 2026.3.24
- —CVE-2026-35663OpenClaw: Gateway Backend Reconnect lets Non-Admin Operator Scopes Self-Claim operator.adminfrom 0, <= 2026.3.24
- from 0, < 2026.3.22
- —CVE-2026-35639OpenClaw Gateway: RCE and Privilege Escalation from operator.pairing to operator.admin via device.pair.approvefrom 0, < 2026.3.22
- from 0, < 2026.3.22
- —CVE-2026-35637OpenClaw: Tlon cite expansion happens before channel and DM authorization is completefrom 0, < 2026.3.22
- —CVE-2026-35626OpenClaw is vulnerable to unauthenticated resource exhaustion through its voice call webhook handlingfrom 0, < 2026.3.22
- —CVE-2026-35633OpenClaw: Remote media error responses could trigger unbounded memory allocation before failurefrom 0, < 2026.3.22
- —CVE-2026-35627OpenClaw: Nostr inbound DMs could trigger unauthenticated crypto work before sender policy enforcementfrom 0, < 2026.3.22
- —CVE-2026-34426OpenClaw: Windows media loaders accepted remote-host file URLs before local path validationfrom 0, < 2026.3.22
- from 0, < 2026.3.23
- —CVE-2026-35634OpenClaw: Gateway Canvas local-direct requests bypass Canvas HTTP and WebSocket authenticationfrom 0, < 2026.3.23
- from 0, < 2026.3.23
- from 0, < 2026.03.28
- from 0, < 2026.2.17
- from 0, < 2026.3.12
- —CVE-2026-32920OpenClaw: Workspace plugin auto-discovery allowed code execution from cloned repositoriesfrom 0, < 2026.3.12
- —CVE-2026-34506OpenClaw's MS Teams sender allowlist bypass when route allowlist is configured and sender allowlist is emptyfrom 0, < 2026.3.8
- —CVE-2026-33574OpenClaw's skills-install-download can be redirected outside the tools root by rebinding the validated base pathfrom 0, < 2026.3.8
- —CVE-2026-22170OpenClaw: BlueBubbles (optional plugin) pairing/allowlist mismatch when allowFrom is emptyfrom 0, < 2026.2.22
- —CVE-2026-31995OpenClaw has Windows Lobster shell fallback command injection in constrained fallback path>= 2026.1.21, < 2026.2.19
- from 0, < 2026.2.22
- —CVE-2026-32050OpenClaw's Signal reaction-only status events could, in limited cases, be enqueued before access checksfrom 0, < 2026.2.25
- —CVE-2026-27523OpenClaw's sandbox bind validation could bypass allowed-root and blocked-path checks via symlink-parent missing-leaf pathsfrom 0, < 2026.2.24
- from 0, < 2026.2.25
- —CVE-2026-31998OpenClaw's Synology Chat dmPolicy=allowlist failed open on empty allowedUserIds, allowing unauthorized agent dispatch>= 2026.2.22, < 2026.2.24
- from 0, < 2026.2.22
- from 0, < 2026.2.21
- —CVE-2026-32063OpenClaw Improperly Neutralizes Line Breaks in systemd Unit Generation Enables Local Command Execution (Linux)from 0, < 2026.2.21
- —CVE-2026-22176OpenClaw has a Command Injection via unescaped environment assignments in Windows Scheduled Task script generationfrom 0, < 2026.2.19
- from 0, < 2026.2.22
- —CVE-2026-32042OpenClaw unpaired device identity can bypass operator pairing and self-assign operator scopes with shared auth>= 2026.2.22, < 2026.2.25
- —CVE-2026-31994OpenClaw Windows Scheduled Task script generation allowed local command injection via unsafe cmd argument handlingfrom 0, < 2026.2.19
- —CVE-2026-22217OpenClaw: shell-env trusted-prefix fallback allowed attacker-controlled binary execution via $SHELL>= 2026.2.22, < 2026.2.23
- —CVE-2026-27670OpenClaw: ZIP extraction race could write outside destination via parent symlink rebindfrom 0, < 2026.3.2
- —CVE-2026-22181OpenClaw's web tools strict URL guard could lose DNS pinning when env proxy is configuredfrom 0, < 2026.3.2
- —CVE-2026-31990OpenClaw: stageSandboxMedia destination symlink traversal can overwrite files outside sandbox workspacefrom 0, < 2026.3.2
- —CVE-2026-32052OpenClaw's system.run shell-wrapper positional argv carriers could execute hidden commands under misleading approval textfrom 0, < 2026.2.24
- from 0, < 2026.2.25
- —CVE-2026-32053OpenClaw's voice-call Twilio webhook replay could bypass manager dedupe because normalized event IDs were randomized per parsefrom 0, < 2026.2.23
- from 0, < 2026.2.21
- from 0, < 2026.2.21
- —CVE-2026-22171OpenClaw vulnerable to path traversal in Feishu media temp-file naming allows writes outside os.tmpdir()from 0, < 2026.2.19
- —CVE-2026-32046OpenClaw: Chrome --no-sandbox disabled OS-level browser sandbox in sandbox browser containerfrom 0, < 2026.2.21
- from 0, < 2026.2.19
- from 0, < 2026.2.23
- from 0, < 2026.2.26
- —CVE-2026-27522OpenClaw: Message action attachment hydration bypasses local media root checks when sandboxRoot is unsetfrom 0, < 2026.2.24
- —CVE-2026-32049OpenClaw's inbound media downloads could exceed configured byte limits before rejection across multiple channelsfrom 0, < 2026.2.22
- —CVE-2026-22175OpenClaw's exec allow-always can be bypassed via unrecognized multiplexer shell wrappers (busybox/toybox sh -c)from 0, < 2026.2.23
- from 0, < 2026.2.25
- —CVE-2026-22178OpenClaw has ReDoS and regex injection via unescaped Feishu mention metadata in RegExp constructionfrom 0, < 2026.2.19
- —CVE-2026-31993OpenClaw macOS companion app (beta): allowlist parsing mismatch for system.run shell chainsfrom 0, < 2026.2.22
- from 0, < 2026.2.21
- —CVE-2026-31997OpenClaw: system.run approvals did not bind PATH-token executable identity, enabling post-approval executable rebindfrom 0, < 2026.3.1
- from 0, < 2026.3.1
- —CVE-2026-31999CpenClaw's ACPX Windows wrapper shell fallback allowed cwd injection in specific paths>= 2026.2.26, < 2026.3.1
- —CVE-2026-32048OpenClaw's sandboxed sessions_spawn now enforces sandbox inheritance for cross-agent spawnsfrom 0, < 2026.3.1
- —CVE-2026-28461OpenClaw has unbounded memory growth in Zalo webhook via query-string key churn (unauthenticated DoS)from 0, < 2026.3.1
- —CVE-2026-4039OpenClaw: Skill env override host env injection via applySkillConfigEnvOverrides (defense-in-depth)from 0, < 2026.2.21
- —CVE-2026-27576OpenClaw: ACP prompt-size checks missing in local stdio bridge could reduce responsiveness with very large inputsfrom 0, < 2026.2.19
- from 0, < 2026.2.19
- from 0, < 2026.2.19
- —CVE-2026-27484OpenClaw Discord moderation authorization used untrusted sender identity in tool-driven flowsfrom 0, < 2026.2.18
- from 0, < 2026.2.15
- —CVE-2026-27007OpenClaw's sandbox config hash sorted primitive arrays and suppressed needed container recreationfrom 0, < 2026.2.15
- from 0, < 2026.2.15
- from 0, < 2026.2.15
- from 0, < 2026.2.15
- from 0, < 2026.2.15
- from 0, < 2026.2.14
- >= 2026.1.8, < 2026.2.14
- from 0, < 2026.2.14
- —CVE-2026-26327OpenClaw allows unauthenticated discovery TXT records to steer routing and TLS pinningfrom 0, < 2026.2.14
- from 0, < 2026.2.14
- >= 2026.2.6-0, < 2026.2.14