The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using pkill and lsof to manage local development processes and verify port availability during the cleanup and smoke-testing phases.\n- [REMOTE_CODE_EXECUTION]: Dynamically generates a JavaScript smoke-testing script (/tmp/dean-smoke.mjs) and executes it using the bun runtime. It also executes react-doctor using npx -y react-doctor@latest, which downloads and runs third-party code at execution time.\n- [EXTERNAL_DOWNLOADS]: Fetches the latest version of the react-doctor package from the npm registry during the Phase 3.6 auditing step.\n- [DATA_EXFILTRATION]: The skill documentation explicitly states that the react-doctor tool 'phones home by default', confirming that project-related data or reports are transmitted to external servers.\n- [PROMPT_INJECTION]: The skill accepts user-supplied change descriptions as arguments, which are used to guide the entire automated process. The skill has high-privilege capabilities such as process management and script execution, but lacks input sanitization or boundary markers.\n
Ingestion points: User-provided arguments in SKILL.md.\n
Boundary markers: None identified.\n
Capability inventory: Shell command execution (pkill, lsof), runtime script generation/execution (bun), and package execution (npx).\n
Sanitization: None identified.

five-phase-pass