The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using the Bash tool to inspect git history via git log. This is a core part of its workflow to identify undocumented changes.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of untrusted repository data.
Ingestion points: Existing documentation (README.md), CHANGELOG.md, git commit messages, and project source code as specified in the workflow steps in SKILL.md.
Boundary markers: Absent. There are no instructions to use delimiters or guardrails to differentiate between repository data and agent instructions.
Capability inventory: Access to Bash for command execution and Write/Edit tools for modifying files across the repository.
Sanitization: Absent. The skill does not perform validation or filtering of the content read from the repository before using it to generate documentation or propose actions.

repo-doku