The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from the repository, creating a potential attack surface.
Ingestion points: The skill reads README.md, wiki/, docs/*.md, CLAUDE.md, AGENTS.md, and various source code directories (e.g., plugin/, commands/, agents/) to reconcile claims.
Boundary markers: The instructions do not specify any delimiters or warnings to the agent for handling potentially malicious instructions embedded in the documentation or code files.
Capability inventory: The skill utilizes rg (ripgrep) and git show for repository exploration, and performs file modification tasks to update documentation based on its findings.
Sanitization: There are no sanitization or input validation measures mentioned to mitigate the risks associated with processing untrusted content.

docs-sync