Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
clawscan
v1.0.0perform first-pass security checks for openclaw deployments by registering the client, checking whether the installed clawscan package is outdated, checking...
⭐ 0· 211·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
The SKILL.md and README describe a cloud-backed scanner that computes SHA-256 hashes of installed skill files, enumerates listening sockets, and posts minimal metadata to https://clawscan.autosec.dev — all coherent with the stated purpose. However the registry metadata provided earlier lists no required binaries or config paths, while SKILL.md metadata and the bundled scripts clearly require python3 and optionally ss/lsof and persist client/schedule files under ~/.openclaw/clawscan. This mismatch between declared registry requirements and the skill's instructions/configuration is an incoherence to address.
Instruction Scope
Instructions are narrowly scoped: collect client_id, openclaw version, relative skill file paths + sha256, and listener metadata; do not upload file contents or secrets by default. That's appropriate for the scanner purpose. Two operational notes: (1) scheduled-scan persists schedule.json and runs silently unless a risk is found — that silent behavior may surprise users and repeatedly transmit hashes/listener metadata without frequent user-visible output; (2) SKILL.md advocates using the bundled scripts and specific commands (openclaw --version, cat package.json | jq ...) which are reasonable but rely on tools (jq) not declared.
Install Mechanism
No install spec (instruction-only) and only two small helper scripts are included. There are no downloads or archive extraction steps. This is low-risk from an install/execution footprint perspective.
Credentials
The skill does not request environment variables or API keys in the manifest, which is consistent with the stated minimization policy. However the API contract references an external service and notes 'Auth scheme is intentionally unspecified' — it's unclear whether an operator-provided bearer token or header is required in practice. Sending client_id, version strings, sha256 hashes, and listener metadata to a remote endpoint is proportional for a remote hash-matching service, but the lack of declared auth and the absence of a privacy/auth disclosure in the skill metadata is a gap.
Persistence & Privilege
The skill instructs persisting a client UUID and schedule state under ~/.openclaw/clawscan (client.json and schedule.json) and supports scheduled scans that are silent when no issues are found. The registry metadata did not declare required config paths. Persisting state and enabling periodic background scans without explicit, visible user notification increases the risk surface and should be an explicit opt-in.
What to consider before installing
What to consider before installing:
- This skill will compute SHA-256 of your installed skill files and gather local listener metadata, then send those minimal records plus an anonymous client UUID to https://clawscan.autosec.dev for server-side matching. That behavior is consistent with a remote hash-matching scanner, but be sure you are comfortable sending that metadata off-host.
- Metadata mismatches: the SKILL.md and scripts require python3 and optionally ss/lsof and persist files under ~/.openclaw/clawscan, but the registry entry did not declare these requirements or config paths. Ask the publisher to correct the manifest to list required binaries and the persisted paths.
- Scheduled scans: scheduled-scan persists schedule.json and runs silently unless a finding occurs. If you prefer visibility, do not enable scheduled scans or require an explicit prompt/consent before each upload.
- Authentication & endpoint: the API contract says auth is unspecified. Confirm whether the public endpoint requires an operator-provided token or whether you can self-host the ClawScan service (and obtain policies/privacy statement) before sending data.
- Quick mitigations: run the helper scripts locally to inspect outputs (python scripts/collect_skill_hashes.py and python scripts/list_listeners.py) before any network call; if you must use remote matching, prefer a self-hosted endpoint or insist on explicit, documented auth and a privacy policy.
- What would change this assessment: if the registry metadata is updated to declare required binaries and config paths, if the publisher documents the auth scheme and privacy policy for the endpoint, and if scheduled scans require explicit user opt-in with visible reporting, the skill would likely be considered benign/coherent.Like a lobster shell, security has layers — review code before you run it.
latestvk973jgaep7m9fa2cr2ehwk8ynx82nrra
License
MIT-0
Free to use, modify, and redistribute. No attribution required.