Skillv1.0.12

ClawScan security

blog-polish-en-astro-cn · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 22, 2026, 11:53 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: This is an instruction-only blog-polishing/translation skill that reads a local draft and writes polished Markdown outputs; its requirements and behavior are coherent aside from minor metadata/path inconsistencies.
Guidance: This skill is primarily an instruction-only workflow that reads a markdown draft from your filesystem and writes polished/translated markdown files. Before installing or running: (1) Confirm the draftPath and outputDir defaults point where you expect (some included files contain a hardcoded /home/jeff path); update them if necessary. (2) Ensure jq is available if your runtime relies on the SKILL metadata (the finalize step calls jq to emit JSON). (3) Do not put secrets or sensitive credentials inside drafts — the content is sent to the LLM for polishing/translation. (4) Because this is an LLM-driven skill, consider which model/provider will be used and review its privacy policy if draft contents are sensitive. Other than the minor metadata/path inconsistencies, the skill's declared purpose matches what it does.

Review Dimensions

Purpose & Capability: okThe name/description (polish + translate a blog draft and produce an Astro-compatible English file) matches the workflow: read a draft, call the LLM to polish/translate, write out markdown files. The only mismatch is a metadata note that the skill 'requires' jq while the registry summary reported no required binaries — jq is actually used in finalize steps to emit JSON and is reasonable for this purpose.
Instruction Scope: noteInstructions explicitly read the draftPath file and write outputs to outputDir (creating directories). This is expected. There is no instruction to read unrelated system files, access credentials, or call external endpoints other than invoking the LLM. Note: some included files reference a hardcoded user path (/home/jeff/...) which is likely a leftover and should be corrected or checked before use.
Install Mechanism: okNo install spec or code files are present (instruction-only), so nothing is downloaded or written during install. Low install risk.
Credentials: okThe skill does not request environment variables, credentials, or config paths. It operates on user-specified file paths. This is proportionate to a local polishing/translation tool. Reminder: the SKILL metadata lists jq as a required tool — not a credential but a binary dependency.
Persistence & Privilege: okalways:false and no modifications to other skills or system-wide settings. The skill creates files under the user-specified outputDir only, which is appropriate for its purpose.