The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from session history and repository artifacts to perform its analysis, creating a standard surface for indirect prompt injection.
Ingestion points: Reads primary session history via /pstack-recall and examines repository files such as package.json, CLAUDE.md, and AGENTS.md as specified in SKILL.md.
Boundary markers: The instructions do not define explicit delimiters or instructions to ignore embedded commands within the analyzed history or repository files.
Capability inventory: The skill identifies candidates for improvement and prepares instruction edits for review; it invokes /domain-modeling and /pstack-swarm for managing changes and delegated work.
Sanitization: There is no evidence of explicit sanitization or filtering of the ingested content before processing.

retro