The Cognition Layer Framework (CLF) is a public specification for building self-aware AI agent systems with explicit orchestration, verification, pattern detection, self-correction, context management, and adapter contracts.
- Version:
1.0.0 - Status:
Stable - License:
MIT
- Cognition-Layer-Framework-Specification.md: normative framework specification
- CONFORMANCE.md: conformance profiles and release-claim guidance
- SECURITY.md: security reporting and supported-version policy
- CHANGELOG.md: release history
- LICENSE: project license
See CONFORMANCE.md for the full suite requirements and public claim rules. The three supported profiles are:
- Core —
CognitionOrchestrator+VerificationGuardian+EventBus+CognitionAdapter - Standard — Core +
PatternDetector+PatternPersistenceCore - Full — Standard +
ContextManager+SelfCorrectionTrigger
Experimental and preview surfaces (Discovery Mode, external invalidation sources, advanced conflict resolution, shared/global pattern persistence) may be implemented but are not part of 1.0 conformance claims.
- A stable public specification
- A contract for interoperable implementations
- A reference point for conformance claims
- A production reference implementation
- A test harness or certification service
- A guarantee that experimental surfaces are frozen
- Read Cognition-Layer-Framework-Specification.md for the normative contract, including TypeScript interface blocks and MUST/SHOULD/MAY conventions.
- Read CONFORMANCE.md to choose a target profile and understand suite requirements.
- Review SECURITY.md before publishing adapters or integrations.
- Use the Migration Path in Section 10 of the specification to plan your implementation order (Adapter → Core → Standard → Full → Experimental).
v1.0.0 is the first stable public release of the CLF specification. The core, standard, and full conformance profiles are defined in the spec, and experimental extensions remain clearly marked. Normative surfaces are identified by TypeScript interface blocks and the use of MUST, SHOULD, and MAY.