Skillv1.0.0

ClawScan security

Self Thinking Agent · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

BenignApr 30, 2026, 1:49 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's code and runtime instructions align with its stated purpose (metacognition / bias detection); it requests no credentials, performs only local text analysis, and has no install or network behavior.
Guidance: This skill appears internally consistent and low-risk: it analyzes text for simple bias indicators using regex heuristics, with no network calls or secret access. Before enabling it for autonomous use, consider: (1) its bias detection is heuristic and will produce false positives/negatives—do not rely on it as sole authority for critical decisions; (2) do not point the analyze command at sensitive files (it reads and prints file contents); and (3) if you will allow the agent to invoke the skill autonomously, review how the agent passes content to the skill so it does not inadvertently process confidential data. If you want higher assurance, run the included JS in a sandboxed environment and test with non-sensitive examples first.

Purpose & Capability: okName, description, SKILL.md, and the included JavaScript all implement the same metacognition / bias-detection purpose. There are no unrelated binaries, env vars, or config paths requested that would be inconsistent with the stated function.
Instruction Scope: okSKILL.md defines pre-output checks, bias-detection steps, and confidence guidance; it does not instruct the agent to access unrelated system state, secrets, or external endpoints. The included CLI code reads only files explicitly passed to its analyze command and otherwise operates on provided reasoning text.
Install Mechanism: okNo install spec or external downloads are present. The skill is instruction-first and includes a local JS file; there are no URLs, package installs, or archive extracts that would write or execute remote code on install.
Credentials: okThe skill requires no environment variables or credentials. Note: its analyze CLI reads local files you give it, so feeding sensitive files to the analyze command would expose their contents to the skill's runtime (but there is no network transmission in the code).
Persistence & Privilege: okalways:false and normal model invocation settings. The skill does not request persistent system-wide privileges or modify other skills' configs.