Skillv1.0.8

ClawScan security

eonik creative audit · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 16, 2026, 11:32 AM

Verdict: Benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's code, instructions, and requested credential (EONIK_API_KEY) are consistent with its stated purpose (running an external eonik audit API and presenting results); there are no obvious mismatches or hidden exfiltration mechanisms in the provided files.
Guidance: This skill appears to do what it says: it calls eonik's API with your EONIK_API_KEY and prints/writes the audit report. Before installing or scheduling it: 1) Verify you trust the eonik.ai service and that the endpoint (https://api.eonik.ai) is expected for your account. 2) Treat output/ directory as sensitive — reports include Meta Ad IDs and thumbnails; ensure DLP/permissions prevent accidental sharing. 3) Use a dedicated, scoped API key for this skill and rotate/revoke if you stop using it. 4) Confirm you want daily automated audits before adding the cron job (scheduling increases the blast radius if the key is compromised). 5) Note the minor metadata mismatch: skill.json requires python3 even though the registry snapshot earlier showed none — ensure Python 3.7+ is available on the runner. If you want extra assurance, review or run the scripts in an isolated environment and inspect network traffic to confirm calls go only to api.eonik.ai.

Review Dimensions

Purpose & Capability: okName/description match behavior: the skill runs a local pipeline which calls the eonik backend (api.eonik.ai) using a single EONIK_API_KEY to produce an audit report. Requiring a single service API key is appropriate for this purpose. The README, SKILL.md, and scripts all align on the advertised functionality.
Instruction Scope: okRuntime instructions are limited to checking/creating config.json, invoking the included pipeline (python3 scripts/pipeline.py --config config.json), and reading the generated output file. The scripts only read the config and write an output JSON containing audit findings (ad IDs, spend, thumbnails). There are no instructions to scan unrelated files, read other credentials, or transmit local files to unexpected endpoints beyond the declared api.eonik.ai endpoint.
Install Mechanism: okThis is an instruction-only skill with included Python scripts (no external install downloads or archive extracts). There is no install spec that fetches arbitrary code at runtime. Dependencies are standard library-only and the repository includes the script files.
Credentials: noteThe skill requests only one credential: EONIK_API_KEY (primaryEnv). That is proportionate to a service that calls a vendor API. One small inconsistency: top-level metadata in the provided registry snapshot said 'required binaries: none', while skill.json lists python3 in 'requirements.bins' — in practice Python is required to run the included scripts. Otherwise no unrelated secrets or extra credentials are requested.
Persistence & Privilege: okThe skill is not always-enabled (always:false) and uses standard invocation. It suggests using OpenCLAW's cron scheduler to run periodically; scheduling itself is a normal convenience but will create regular scans and outputs, so users should enable scheduling intentionally. The skill does not modify other skills or global config.