smart-git-commit
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses multiple git commands to automate the development workflow, including
git add,git commit, andgit push. These provide the agent with broad access to the local file system and configured remote repositories.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the repository, creating a surface for indirect prompt injection attacks.\n - Ingestion points: Untrusted data enters the agent context via
git status,git diff, andgit log(SKILL.md), which read file changes and history.\n - Boundary markers: The workflow lacks explicit boundary markers or instructions to treat data read from the repository as potentially untrusted instructions.\n
- Capability inventory: The skill has the capability to write to the repository and transmit data to a remote server via
git commitandgit push.\n - Sanitization: No sanitization or escaping of file content is performed before the agent uses it to generate commit messages or categorize changes.\n- [PRIVILEGE_ESCALATION]: The skill contains explicit instructions to "Run the push outside the sandbox" (SKILL.md). This is a directive to bypass the security environment provided by the agent platform to gain unrestricted network access, which is considered a privilege escalation pattern.
Audit Metadata