What problem does it solve?
Curates and structures the information fed to AI assistants to prevent hallucinations, reduce irrelevant output, and ensure adherence to project conventions when working on code or specs.
Core Features & Use Cases
- Rules Files & Persistent Context: Establishes project-wide rules files (CLAUDE.md, .github/copilot-instructions.md, etc.) so agents always follow team conventions and constraints.
- Selective Context Loading: Guides which spec sections, source files, and test outputs to include per task to keep focus and avoid context flooding.
- Conversation & Confusion Management: Provides patterns for summarizing long histories, surfacing conflicts between spec and code, and emitting inline plans before execution.
- Use Case: Onboarding a new feature, load the rules file, the feature's spec excerpt, example implementations, and tests to let the agent implement a change without inventing APIs.
Quick Start
Load the project's rules file, the relevant spec section, and the specific source and test files for the task, then ask the agent to propose a short implementation plan.