commit-craft
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingest and processes output from
git diffandgit status, which are untrusted data sources that could contain adversarial instructions within file content. - Ingestion points: Found in
SKILL.mdwhere the agent is instructed to usegit statusandgit diffto understand changes. - Boundary markers: The skill does not provide clear delimiters to isolate the diff data from instructions.
- Capability inventory: The skill executes shell-based git commands, including
addandcommit. - Sanitization: While it includes a safety sweep to prevent committing secrets like AWS keys or private keys, it does not include sanitization against prompt injection patterns in the data it reads.
- [COMMAND_EXECUTION]: The skill utilizes a bash heredoc construction for executing
git commitcommands. This pattern is susceptible to command injection if the agent generates a commit message that includes shell control characters or the heredoc delimiter. - [EXTERNAL_DOWNLOADS]: The skill references the author's external GitHub repository for promotional purposes. These links point to a well-known service and the vendor's own project assets.
Audit Metadata