release
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill involves reading external, potentially untrusted data which could contain instructions meant to subvert the agent.
- Ingestion points: The agent reads git commit messages (via
origin/main..origin/dev) and GitHub issue content (labeledon-dev) as specified in the release workflow. - Boundary markers: No delimiters or instructions to ignore embedded commands are present when the agent is instructed to process these external strings.
- Capability inventory: The skill allows for file system modifications (version bumping), git history changes (tagging and pushing), and publishing packages to external registries (npm).
- Sanitization: There is no evidence of sanitization or filtering for the strings retrieved from the git history or the GitHub issue tracker.
- [COMMAND_EXECUTION]: The instructions direct the agent to use standard development CLI tools to manage the software lifecycle.
- Evidence: The skill utilizes
git,bun,npm, andlipoto perform its primary function of automated software releasing.
Audit Metadata