writing-commit-messages
Pass
Audited by Gen Agent Trust Hub on Aug 2, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes un-sanitized output from 'git diff' or 'jj diff'. These diffs represent arbitrary code changes that may contain adversarial instructions in comments or literals designed to influence the agent's behavior. Ingestion points: Workflow instruction to run a diff in SKILL.md. Boundary markers: Absent. Capability inventory: Git and Jujutsu (jj) CLI execution. Sanitization: Absent.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands ('git' or 'jj') to read repository state and apply commits. Evidence: Workflow steps 'Run a diff' and 'Apply the commit'.
Audit Metadata