Skillv1.8.0

ClawScan security

Isfj Coach · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

ReviewApr 17, 2026, 10:41 AM

Verdict: Review
Confidence: medium
Model: gpt-5-mini
Summary: The skill's runtime instructions look coherent for an ISFJ coaching assistant, but repository/metadata and supporting files contain multiple copy‑paste mismatches and unexplained references to an INTP coach and unrelated repo, so the package's provenance and data‑handling expectations are unclear.
Guidance: This skill's instructions for coaching are coherent, but multiple supporting files and metadata do not match (README and keywords reference an INTP coach, repository URL is unrelated, version fields differ). Before installing or enabling persistence: - Ask the publisher/owner to explain and correct the mismatched README, clawhub.json, and versioning so provenance is clear. - Ask where '成长记录' (user records) are stored, how long they're kept, and who can access them. If the platform persists data for skills, confirm retention and deletion options. - Prefer skills with a published homepage or repository that matches the package metadata and a clear privacy note. If you cannot verify the author or repo, avoid enabling persistent storage or sharing sensitive personal details in the skill. - If you decide to try it, test in a low‑risk conversation (no personal identifiers, no sensitive mental‑health disclosures) and confirm the skill retrieves/uses stored records as described. Request an updated release with corrected metadata and an explicit privacy/storage statement.

Review Dimensions

Purpose & Capability: concernThe SKILL.md describes an ISFJ Coach and contains reasonable coaching logic. However, README.md and many metadata fields (clawhub.json keywords/tags, README text, references folder names) are about an INTP coach, the repository URL points to an unrelated repo, and SKILL.md header/version differs from registry version. These inconsistencies indicate sloppy packaging or copy‑paste errors and make it unclear which files are authoritative or whether the skill was repurposed without updating provenance.
Instruction Scope: noteThe instructions are largely self‑contained coaching heuristics and control logic (mode switching, question frames, response length limits). They instruct the agent to 'record' user profiles, actions and insights for later retrieval; no storage API, path, or explicit external endpoint is declared. That behavior is plausible for a coaching skill (persistent user records), but the skill did not declare where or how persistence occurs, which should be clarified for privacy expectations.
Install Mechanism: okInstruction‑only skill with no install spec and no binaries or downloads. This is the lowest install risk and matches an authoring style that only provides runtime directives.
Credentials: okThe skill declares no required environment variables, credentials, or config paths. That aligns with the stated functionality (coaching conversation). The only minor concern is the implicit requirement to persist user data (records) despite no declared storage or consent flow.
Persistence & Privilege: notealways:false (normal). The SKILL.md explicitly instructs storing user profiles and action logs across sessions. That is reasonable for a coach but the skill does not document where data is stored, retention policy, or privacy safeguards. Combined with mismatched metadata, this raises concerns about provenance and data handling rather than privileged system access.