The Agent Skills Directory

[COMMAND_EXECUTION]: The artifact.mjs script utilizes node:child_process.execFile to run the git binary, fetching repository metadata such as the origin URL, current branch, and HEAD SHA to include in the front matter of generated artifacts.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process and act upon review comments stored within plan files, creating a surface for indirect instructions.
Ingestion points: The agent reads plan files using the Read tool and extracts comments using the node scripts/artifact.mjs list command (defined in SKILL.md and artifact.mjs).
Boundary markers: Review comments are delimited by <comment> tags within an  block, which the script parses.
Capability inventory: The skill allows the agent to use Bash, Read, Write, Grep, and Glob tools, alongside the provided management script to modify plan artifacts and project structure.
Sanitization: The artifact.mjs script includes logic to prevent comments from breaking the internal file structure (e.g., forbidding </comment> or <!-- inside attribute values), but it does not sanitize the natural language content of the comments before the agent processes them.

iterate-plan