Back to skill
Skillv1.0.2
ClawScan security
code-doc-guid · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignMar 7, 2026, 6:47 AM
- Verdict
- benign
- Confidence
- medium
- Model
- gpt-5-mini
- Summary
- The skill's requested actions and included script are coherent with a local codebase indexing/navigation tool; it does not ask for credentials or external installs, but it will read and write a local .trae database and generate reports so you should review the script and be prepared to allow filesystem writes.
- Guidance
- This skill appears to be a local code-indexer and its requirements align with that purpose. Before allowing it to run: 1) Review the full scripts/codebase_manager.py for any subprocess/network calls (the visible portion imports subprocess but the file was truncated in the supplied content). 2) Expect the skill to create/modify .trae/ and a local SQLite DB (codebase.db) in your project — back up any important files if concerned. 3) If you run it, prefer doing so in a sandbox or a copy of the repository first. 4) If you do not want a tool writing runtime artifacts to the repo, decline or run it with a separate working directory. If you want higher assurance, provide the remaining portion of scripts/codebase_manager.py so it can be fully audited (particularly to confirm no network exfiltration or arbitrary shell execution).
Review Dimensions
- Purpose & Capability
- okName/description match the observed behavior: an on-disk SQLite-based code index, search, inspect, graph, and update commands. No unrelated environment variables, binaries, or external services are requested.
- Instruction Scope
- noteSKILL.md prescribes that the agent MUST run this skill before any code modification and after modifications. This is consistent with a code-indexing workflow, but it is prescriptive and enforces a blocking step that can affect agent flows. The instructions operate on project files and expect reading many files and generating .trae artifacts—this behaviour is within the stated purpose.
- Install Mechanism
- okInstruction-only skill (no install spec). A Python script is bundled; no external downloads or package installs are requested.
- Credentials
- okNo environment variables, credentials, or unrelated config paths are requested. The script will create/modify local files (.trae/, codebase.db) and expects filesystem write access — that's proportionate to a local indexer.
- Persistence & Privilege
- okSkill is not always-enabled and does not request elevated platform privileges. It writes to project-local paths (e.g., .trae/) and manages its own DB; that is normal for this type of tool. The skill can be invoked autonomously (platform default) but that is not unusual by itself.
