The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a companion Python utility (ap.py) that executes shell commands via subprocess to manage the project repository and lifecycle. This includes initializing git repositories, committing changes, moving files, and launching a local-only HTTP server on 127.0.0.1 to display the run dashboard.
[EXTERNAL_DOWNLOADS]: The ap.py tool performs a network request to the author's public GitHub repository (nick-vels/skills) to check for available updates to the skill definition. This is a legitimate vendor-related operation.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process user-provided project briefs which are then used as instructions for code generation subagents. This introduces an attack surface where a brief could contain hidden malicious instructions. The skill mitigates this through a mandatory 'Redaction gate' that identifies and replaces sensitive API keys or credentials, and a 'Blind Acceptance' phase where an independent agent verifies the build output against the original brief without seeing intermediate plans or specifications.

autopilot