problem-analysis

[Skill Scanner] Instruction directing agent to run/execute external content The manifest instructs immediate execution of an unprovided local Python module and explicitly tells the agent not to inspect before running. While the stated goal (root-cause analysis) can be legitimate, the anti-review imperative and absence of the module or safety constraints create a significant supply-chain risk. Treat this as SUSPICIOUS: do not run the invoked script until its source code has been reviewed, expected resource accesses (files, network endpoints) are declared, and execution is performed under least privilege or within a controlled sandbox. LLM verification: SKILL.md itself is not executable malware but is a high-risk manifest because it instructs immediate execution of an unvetted local Python module and explicitly forbids prior exploration. This pattern enables arbitrary code execution and potential data exfiltration or system modification depending on the referenced script's contents. Do not run the invoked script until its source, provenance, and expected behavior are reviewed and execution is constrained by sandboxing and least-privilege contro