The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands including git log, git diff, ls, and sed to perform repository analysis. It uses values retrieved from a local state file (_/.last-update-docs) as parameters in these commands, which is a common pattern for documentation tools but requires the source file to remain untampered.
[PROMPT_INJECTION]: The skill ingests and processes untrusted data from git commit messages and diffs, creating an indirect prompt injection surface.
Ingestion points: The agent reads the output of git log and git diff into its context during Step 2 of the workflow.
Boundary markers: Absent; there are no delimiters or specific instructions to help the model distinguish between legitimate commit history and potentially malicious instructions embedded in a commit message.
Capability inventory: The skill possesses capabilities to write to local files (CLAUDE.md and _/.last-update-docs) and execute subsequent shell commands based on its analysis.
Sanitization: No sanitization, escaping, or validation of the commit messages or diff content is performed before the agent processes them.

update-docs