pflow-commit

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes multiple bash scripts (git-commit-context.sh, git-commit-push.sh) that perform significant system and repository operations including git add -A, git commit, and git push.
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads untrusted data from the local repository (file contents and diffs) and uses this as the sole input to generate a commit message. An attacker who can influence file content in the repository could potentially embed instructions that influence the agent's behavior during the commit message generation step.
  • Ingestion points: The git-commit-context.sh script reads the content of added files and the diffs of modified files in the working tree.
  • Boundary markers: None identified in the prompt instructions to distinguish between the diff data and the instructions for message composition.
  • Capability inventory: The skill has the capability to perform shell commands, modify the git index, create commits, and push code to remote repositories via git-commit-push.sh.
  • Sanitization: There is no sanitization of the input diff/content before it is processed by the LLM to compose the commit message.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 08:18 PM
Security Audit — agent-trust-hub — pflow-commit