Context Library
The Context Library is the knowledge base that grounds AI agents in your codebase conventions, domain knowledge, and standards. Context quality is the single biggest driver of agent output quality.
What goes in the Context Library
Architecture Decision Records (ADRs) — Why key decisions were made. Agents use these to avoid re-litigating settled decisions.
Coding Standards — Patterns, conventions, and anti-patterns for your tech stack. Include naming, structure, and testing expectations.
API Contracts — OpenAPI specs, GraphQL schemas, or API documentation. Agents use these to generate correct integration code.
Personas — User research summaries. Agents use these to assess whether proposed solutions match user needs.
Compliance Controls — Relevant regulatory requirements (GDPR data handling rules, FCA obligations, etc.). Security and Compliance agents reference these.
Domain Glossary — Business terminology unique to your domain. Reduces misunderstanding in generated code and documentation.
Adding entries
Click Add Entry. Paste or write the content. Tag with relevant agent types (e.g. an API contract is most relevant to Coder and Tester agents). Set visibility: All Agents, Selected Types, or Manual Selection per run.
Quality guidelines
- Keep entries concise — agents are better with specific, accurate context than vague, comprehensive documents.
- Prefer concrete examples over abstract rules: "Use `const` unless reassignment is needed" beats "use appropriate const/let".
- Review and update entries quarterly — stale context misleads agents as much as no context does.
- Start with 3–5 high-quality entries rather than bulk-importing everything.