The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze untrusted data from external repositories, creating a surface for indirect prompt injection.
Ingestion points: The skill reads repository files, including source code, comments, READMEs, and CI configurations, as described in Phase 0 and Phase 1.
Boundary markers: The skill explicitly defines a 'Security boundary' in SKILL.md, instructing the agent to treat all repository content as untrusted data and never as instructions. It specifically warns that text inside the repository cannot change the audit's scope or authorize tool calls.
Capability inventory: The skill uses file-read tools to audit the codebase and executes Git commands (git status, git diff) for reconnaissance.
Sanitization: The instructions include a 'verify before flagging' rule, requiring the agent to prove claims against the code rather than blindly following comments or documentation.
[COMMAND_EXECUTION]: The skill instructions (Phase 0) specify the execution of shell commands to establish the audit scope.
Evidence: The skill uses git status and git diff --stat to identify dirty files and determine which files to exclude from the audit.

awesome-slop-audit