Back to skill
Skillv1.0.0
ClawScan security
Agent Workflow Enforcer · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
SuspiciousMar 1, 2026, 3:35 AM
- Verdict
- suspicious
- Confidence
- medium
- Model
- gpt-5-mini
- Summary
- The skill's purpose and the included gate script are coherent and low-risk, but the runtime instructions reference additional scripts and files that are not present and the source/homepage look like placeholders — these inconsistencies merit caution before installing.
- Guidance
- This skill appears to do what it says: gate checks and enforcing output-format conventions. However: - SKILL.md references create_style_context.py and detect_learning.py but only scripts/gate.py is included. Ask the publisher for the missing scripts or inspect them before running anything that would create or learn from files. - The homepage/source points to an example GitHub URL (github.com/example/...), and the package author is anonymous; prefer code from a real, trusted repo or request provenance. - The gate script itself is harmless (prints checklists) but the missing scripts could perform file writes or network activity — validate their contents and run in a sandbox first. Practical next steps before installing/using: 1. Request the full repository or the missing script files and review them for file I/O and network calls. 2. Test locally in an isolated environment (or container) to observe any file writes (style-context.yaml, learnings.jsonl) and ensure no unexpected network requests occur. 3. If you will add the suggested snippets to AGENTS.md or system prompts, do so intentionally and keep backups of existing prompts/configs. If you can't obtain the missing scripts and provenance, treat the package as incomplete and avoid running anything that would modify agent/system files.
Review Dimensions
- Purpose & Capability
- okName/description (enforcing agent workflows with gates, output format, and style persistence) match the included gate.py and the SKILL.md instructions. No unrelated credentials, binaries, or installs are requested.
- Instruction Scope
- concernSKILL.md instructs the agent to run several scripts (gate.py, create_style_context.py, detect_learning.py) and to add enforced blocks to AGENTS.md/system prompts. Only gate.py is present in the package; create_style_context.py and detect_learning.py are referenced but not included. The instructions also ask agents to read/write local files (style-context.yaml, learnings.jsonl) and to modify AGENTS.md/system prompts — these actions are expected for the stated purpose but you should confirm the missing scripts and file-write behavior before use.
- Install Mechanism
- okNo install spec is provided and the skill is instruction-first with one small Python script. Nothing is downloaded or installed by the package itself.
- Credentials
- okThe skill declares no required environment variables, credentials, or config paths. The gate.py script only prints checklists and does not read secrets or network resources.
- Persistence & Privilege
- okFlags show always=false and the skill is user-invocable; it does not request permanent agent presence. The SKILL.md recommends editing AGENTS.md / system prompts, which is a user action and not a stealthy privilege request.