The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Bash to execute git commands for repository analysis. It includes advanced defensive measures to prevent argument injection, such as consistently using the -- and --end-of-options separators to handle branch names that may start with dashes or contain shell metacharacters.\n- [DATA_EXPOSURE]: The skill accesses branch history and commit metadata to perform its analysis. It uses git fetch to update remote-tracking branches from the project's configured remotes, which is standard behavior for git-based development tools.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from git commit messages.\n
Ingestion points: Commit messages are ingested through git log and git cherry commands in the analysis script.\n
Boundary markers: Data is presented to the user in structured markdown tables, and the skill requires human review of the plan before outputting a command.\n
Capability inventory: The skill is limited to analysis and string generation; the execution of the final git rebase command is strictly reserved for the human developer, mitigating the risk of autonomous destructive actions.\n
Sanitization: The analysis script implements dedicated json_escape, sh_quote, and utf8_sanitize functions to verify and clean commit messages and branch names, preventing them from interfering with the JSON structure or causing shell command injection.

smart-rebase