The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several git and grep commands to resolve milestone history and structure. These are used for the primary purpose of milestone assessment.
[PROMPT_INJECTION]: The skill explicitly includes a section regarding untrusted input. It instructs the agent to treat data from the roadmap and human rationales as passive data and never obey instructions embedded within them. It also enforces strict regex shape checks (^MILE-[0-9]+$ and ^[0-9a-f]{40}$) for inputs before they reach the shell.
[DATA_EXPOSURE]: The skill reads from local roadmap and spec files (docs/roadmap/INDEX.md, docs/specs/INDEX.md, docs/product/vision.md). It does not perform network operations or access sensitive credentials outside of the project scope.
[REMOTE_CODE_EXECUTION]: No remote code execution patterns or external package installations were detected. The skill relies on local environment tools (git, grep, sed).

assess-milestone