What problem does it solve?
Engineering decisions like "should we adopt this library?", "what do you think of this design doc?", or "which of these approaches fits our project?" often get answered with generic advice instead of evidence grounded in the actual codebase. This Skill forms a decisive, project-grounded point of view by verifying claims against the repository, prior decisions, and external sources before issuing a graded verdict.
Core Features & Use Cases
- Graded adoption verdicts: Evaluates external tools, libraries, or practices against the project's incumbents, call-sites, and constraints, scaled by a three-level reversibility tier.
- Document takes and approach-set positions: Delivers a holistic judgment on a supplied document or a decision among supplied options, distinct from findings-style review.
- Cross-model panel (oracle): Optionally consults peer models (Codex, Claude, Grok, Cursor, Composer) through read-only sandboxed routes, reconciles disagreement, and discloses per-peer positions.
- Use Case: You paste a link to a new authentication library and ask whether to adopt it. The Skill scouts your codebase for the current auth implementation, checks external evidence, scans prior decision records, and returns a graded Adopt/Trial/Hold/Reject verdict with reasoning.
Quick Start
Ask the agent to give a point of view on whether your project should adopt a specific tool, migrate to a new approach, or which of your listed options it should choose.
