The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill dynamically assembles and executes Python snippets for git-filter-repo and shell scripts for git-filter-branch. These scripts interpolate user-provided values for name and email, which could result in command or code injection if the agent does not strictly sanitize the input before shell execution.
[COMMAND_EXECUTION]: The skill makes extensive use of system command execution via git, gh, glab, and curl to perform repository operations, verify identities, and interact with host-specific APIs.
[EXTERNAL_DOWNLOADS]: The instructions recommend installing the git-filter-repo tool via multiple package managers such as pip, brew, apt, dnf, and scoop. This involves downloading and executing third-party code from external registries.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from untrusted git repositories, including commit history and metadata. 1. Ingestion points: Untrusted data enters the agent context via git clone, git log, and git show as described in Phase 0. 2. Boundary markers: The skill uses specific git format strings (e.g., --format='%H %ae %ce') to delimit data, which provides some structure but is not a complete defense against adversarial content. 3. Capability inventory: The skill possesses significant capabilities, including git push --force, git remote manipulation, and git config modifications. 4. Sanitization: There are no explicit instructions for sanitizing repository content or commit messages before processing or display.

awesome-git-author-rewrite