The Agent Skills Directory

[SAFE]: The skill's primary function is to audit and reconcile documentation files for consistency. It demonstrates secure interaction patterns by requiring user confirmation for fixes unless a fix argument is explicitly provided by the user.\n- [DATA_EXFILTRATION]: References to local configuration paths such as ~/.shep/repos/ and the use of specific port numbers like 3030 are specific to the Shep SDLC platform's documented functionality and do not indicate unauthorized data access.\n- [PROMPT_INJECTION]: The skill processes documentation files which serve as ingestion points for potentially untrusted data.\n
Ingestion points: Reads repository documentation such as README.md, CLAUDE.md, and markdown files under the docs/ directory.\n
Boundary markers: No explicit boundary markers or "ignore instructions" wrappers are defined for the documentation content during processing.\n
Capability inventory: The skill utilizes the Task tool for subagent orchestration and the Edit tool for applying file changes to the codebase.\n
Sanitization: The skill includes a dedicated "Safety Guidelines for Fixes" section instructing the agent to perform minimal, targeted edits, preserve existing style, and avoid adding new content or features beyond the scope of the fix.\n- [EXTERNAL_DOWNLOADS]: The skill contains links to https://shep.bot and the repository https://github.com/shep-ai/shep. These are official resources belonging to the vendor (Shep AI) and are consistent with the skill's identity and metadata.

cross-validate-artifacts