The Agent Skills Directory

[Indirect Prompt Injection Surface]: The skill is designed to ingest and analyze git blame history from the repository to identify code contributions. This represents a potential surface where untrusted data (git commit messages or author names in the repository) could influence the agent's behavior during the analysis phase.
Ingestion points: Git blame history and repository file paths (SKILL.md).
Boundary markers: The prompt lacks explicit delimiters or instructions for the agent to ignore potentially malicious content embedded in git history.
Capability inventory: The skill is intended to generate entries for and modify the .github/CODENOTIFY file.
Sanitization: There are no explicit sanitization steps mentioned for the data retrieved from git history before it is processed.
**[Data Access
Repository Metadata]:** The skill requests user-provided information such as GitHub handles and alternative email addresses found in git history. While this is used for the legitimate purpose of identifying contributions, it involves the processing of potential PII (email addresses) already present in the repository's metadata.

codenotify