The Agent Skills Directory

[SAFE]: The skill implements a set of common developer verification steps using standard CLI tools like npm, git, grep, and tsc. These operations are local to the development environment and align with standard software engineering practices.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local tool outputs. Ingestion points: Output from build, lint, and test tools (SKILL.md). Boundary markers: None explicitly present for delimiter-based isolation. Capability inventory: Shell execution of npm, npx, pyright, ruff, grep, and git commands (SKILL.md). Sanitization: Logs are truncated using head and tail commands, limiting the data volume processed by the agent.

verification-loop