Skillv1.0.0

ClawScan security

GitHub Repo Teardown · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

BenignMar 15, 2026, 12:18 PM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The instructions, required capabilities, and outputs are coherent with the stated purpose of producing a teardown of a public GitHub repository; nothing requested is disproportionate or unrelated.
Guidance: This skill appears to do what it says: it will make web requests to public GitHub pages, raw file URLs, docs, and discussion/issue pages and then write an HTML report to /mnt/user-data/outputs/. Before using it: only provide public repo URLs (do not paste private repos or tokens), review generated output before sharing (it may include contributor names, commit messages, or other public metadata), and be aware the agent will make outbound web requests (which is normal for this task). If you need analysis of a private repo, confirm how credentials/access should be provided and whether the skill is allowed to use them — this skill does not request or manage such credentials.

Purpose & Capability: okThe skill declares a teardown of GitHub open-source projects and its runtime instructions only use web fetch/search of the repository page, docs, raw source files, issues, and comparable repos — all reasonable and expected for this purpose. No unrelated credentials, binaries, or installs are requested.
Instruction Scope: okSKILL.md restricts activity to fetching public GitHub pages, docs, raw files, and community signals and to producing an HTML report saved to /mnt/user-data/outputs/. It does not instruct reading local system files, environment secrets, or contacting unexpected endpoints beyond listed web resources.
Install Mechanism: okNo install spec or code is included (instruction-only). This is the lowest-risk model and matches the skill's described behavior.
Credentials: okThe skill requires no environment variables, credentials, or config paths. Requested filesystem output (/mnt/user-data/outputs/) is reasonable for saving generated reports.
Persistence & Privilege: okalways is false and model invocation is the platform default. The skill does not request persistent or elevated privileges, nor does it modify other skills or system settings.