ce-commit

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes dynamic context injection (using the !command syntax) to execute several git commands (git status, git diff, git branch, git log, git rev-parse) at load time. These commands are used to pre-populate the agent's context with repository information.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from the repository's working tree and history.
  • Ingestion points: Output from git diff HEAD and git log --oneline -10 in SKILL.md is provided to the agent.
  • Boundary markers: None; the repository content is included directly without delimiters or instructions to ignore embedded commands.
  • Capability inventory: The skill has the ability to execute shell commands (git commit, git checkout, gh repo view), interact with the user, and stage files.
  • Sanitization: No validation or filtering is applied to the data retrieved from the git repository before it is processed by the language model.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 09:14 PM
Security Audit — agent-trust-hub — ce-commit