The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted external data including git branch names and staged code diffs. There is a risk of indirect prompt injection if these sources contain adversarial instructions designed to influence the agent's logic during commit message composition.
Ingestion points: Branch names retrieved via git rev-parse and staged changes from git diff (SKILL.md).
Boundary markers: Absent. The skill does not define delimiters or clear instructions to isolate untrusted content from the surrounding context.
Capability inventory: Execution of shell commands via git and gh tools.
Sanitization: Absent. There are no instructions to escape shell metacharacters in branch names or ticket IDs before they are interpolated into git commit -m arguments.
[COMMAND_EXECUTION]: The skill generates shell commands for git commit by interpolating data from branch names (e.g., ticket IDs) into the command string. If a branch name is maliciously crafted to include shell metacharacters (e.g., backticks or dollar signs), it could lead to arbitrary command execution when the agent runs the generated git commit command.

create-commit