The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user-provided documentation, RFCs, and PR descriptions for rewriting tasks. * Ingestion points: Untrusted text is processed through file contents and metadata as defined in the Contract in SKILL.md. * Boundary markers: Lacks explicit delimiters or instructions to ignore embedded instructions within user input. * Capability inventory: Limited to file modifications and Git metadata updates; no network or subprocess capabilities detected. * Sanitization: No input validation or sanitization of prose is performed.

technical-writing