commit
Pass
Audited by Gen Agent Trust Hub on Mar 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes several local git commands to manage the repository state.
- Evidence: Usage of
git status --short,git diff --stat,git log --oneline -3,git add, andgit commit -m "..."inSKILL.md. - Context: These are standard operations for the skill's primary purpose of committing work.
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) as it processes data that could be controlled by an external actor (e.g., code in files or previous commit messages).
- Ingestion points: File contents and git logs accessed via
git diffandgit logas specified inSKILL.mdinstructions. - Boundary markers: Absent; there are no delimiters or instructions to treat the ingested data as untrusted text.
- Capability inventory: Subprocess execution of git commands which can modify the repository state.
- Sanitization: Absent; the skill does not filter or sanitize the output of git commands before using it to inform subsequent actions.
Audit Metadata