implement
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the execution of a local project script
node scripts/validate-project.mjsat several points in the workflow to verify project state integrity.\n- [EXTERNAL_DOWNLOADS]: The skill performs network queries to official package registries including NPM, crates.io, and GitHub releases to resolve and pin dependency versions.\n- [PROMPT_INJECTION]: The skill processes external, potentially untrusted content from backlog items and wiki concepts, creating a surface for indirect prompt injection.\n - Ingestion points: Backlog records located in
docs/backlog/and wiki concepts indocs/wiki/.\n - Boundary markers: The instructions tell the agent to follow the defined outcome and criteria but do not provide explicit structural delimiters for the ingested data.\n
- Capability inventory: Shell command execution via Node.js, Git operations (commit, merge, branch), and network API requests.\n
- Sanitization: Employs mandatory project validation via scripts and requires manual inspection of staged diffs before integration.
Audit Metadata