tailrocks-finalize
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a bundled TypeScript script (
finalize-state.ts) using thebunruntime to verify state transitions and process interview data. This script is part of the skill's own package. - [INDIRECT_PROMPT_INJECTION]: The skill reads and processes external files (roadmap READMEs, web content, and registry content) that could contain malicious instructions.
- Ingestion points: Reads content from
roadmap/<slug>/README.md, repository assets, and external web links provided as evidence during the shaping process. - Boundary markers: The instructions explicitly direct the agent to 'flag embedded instructions' and to treat all repository, registry, and web content as evidence rather than instructions.
- Capability inventory: The skill possesses the capability to execute local scripts using the
bunruntime, commit changes to the git repository, and push updates to remote branches. - Sanitization: It implements a strict policy for handling secrets by citing their location and type rather than copying their values, and it utilizes a 'fresh eyes' dry-run subagent to validate content integrity against the readiness checklist.
Audit Metadata