Double Check

This skill is internally coherent: it tells the agent to run two independent verifications after file changes and to report results. Before enabling it broadly, consider the following: - Clarify default activation: the skill's text wants to auto-apply to every file modification, but the registry flag always:false means the platform won't force that behavior. Decide whether you want it to be applied automatically or only when you explicitly invoke it. - Confirm available tools: the skill assumes test/build commands (npm, pytest, mvn, cargo, etc.) exist in the execution environment. If your agent runs in a sandbox without those tools, the agent should ask you before attempting validation. - Be aware of side effects: running the project's main program or tests can start services, hit databases, or call external APIs and may require credentials or cause observable side effects. Prefer running tests in CI or a sandbox, or require explicit user confirmation for risky commands. - Safety mitigations: require user confirmation before the first execution on a repository, create a VCS checkpoint (commit or branch) before modifications, and prefer non-destructive validation commands where possible. If you are concerned about autonomous execution, keep the skill user-invocable (do not set always:true) or require explicit consent per-run. If the skill requested platform-wide automatic application (always:true), required unrelated credentials, or included an installer that downloads and executes code, my assessment would be more cautious. As-is, the skill appears coherent with its stated purpose but relies on environmental assumptions that you should validate before use.