git-commit-workflow
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on the execution of several shell commands, including
git status,git add,git diff, andgit commitas defined inSKILL.md. These commands allow the agent to interact with and modify the local filesystem and repository state. - [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it ingests and processes untrusted data from file changes via
git diff --stagedinSKILL.md. Malicious instructions embedded in a project's code could attempt to influence the generated commit message or bypass confirmation steps. - Ingestion points: The output of the
git diff --stagedcommand is processed by the AI inSKILL.mdto generate structured commit messages. - Boundary markers: The instructions do not define specific delimiters (e.g., XML tags or unique markers) to separate the untrusted diff content from the agent's internal logic.
- Capability inventory: The skill possesses capabilities to stage files (
git add) and commit changes (git commit) as seen inSKILL.md. - Sanitization: There are no explicit instructions for the AI to sanitize or ignore embedded instructions within the ingested file content.
Audit Metadata