The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various git commands and shell scripts based on repository inputs provided by the user or discovered by the agent.
[INDIRECT_PROMPT_INJECTION]: The skill provides a mechanism to ingest external data (git repositories). This creates an attack surface where a malicious repository could contain instructions designed to influence the agent's behavior during subsequent analysis of the cloned files.
Ingestion points: checkout.sh clones remote repositories into ~/.cache/checkouts/ based on user-provided or discovered URLs.
Boundary markers: None identified in the provided scripts to distinguish between instructions and data within the cloned repositories.
Capability inventory: The skill uses git clone, git fetch, and git merge via subprocess equivalents in bash. It also writes a timestamp to a tracking file (.git/librarian-last-fetch).
Sanitization: The parse_repo function performs basic parsing and trimming of the input URL/reference but does not sanitize the content of the files once they are cloned to the local system.

librarian