The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the git repository context which could be manipulated by third parties to influence agent behavior.
Ingestion points: The skill reads output from git log --oneline -10 and git diff --cached in SKILL.md to determine the current state and suggest commit messages.
Boundary markers: Absent. There are no instructions for the agent to distinguish between its own system instructions and instructions that might be contained within commit messages or file changes.
Capability inventory: The skill allows for local filesystem modification and network communication via git push/git pull and shell command execution.
Sanitization: Absent. The data retrieved from Git is used directly to guide the agent's logic without filtering.
[COMMAND_EXECUTION]: The skill relies on the execution of multiple shell commands to manage the Git workflow.
Evidence: SKILL.md includes commands for status checking, logging, branching, rebasing, and history modification (e.g., git status, git rebase, git reset).

git-workflow