wk-renovate
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell-based Git and package manager commands to automate branch management and lockfile updates.
- Evidence:
SKILL.mdcontains instructions for the agent to rungit cherry-pick,git push, andgh pr listto manage the lifecycle of the pull requests. - Evidence: Step 3 and 4 in
SKILL.mdinvoke tools likenpm,yarn,pnpm,bundle,cargo, andpoetryto resolve dependencies and verify the repository state. - [EXTERNAL_DOWNLOADS]: The skill facilitates downloads from official package registries when resolving dependency conflicts.
- Evidence: The tool uses standard commands like
npm installandcargo updatewhich interact with well-known official services such as the npm registry and crates.io. - [PROMPT_INJECTION]: The skill processes external data from GitHub pull requests, creating a potential surface for indirect prompt injection.
- Evidence: In Step 1 of
SKILL.md, the agent reads title and body fields from open pull requests via the GitHub CLI. - Mitigation: The skill includes an explicit pause for confirmation in Step 1, requiring the user to review the extracted dependency information and provide approval before the tool performs any file edits or branch operations.
Audit Metadata