Expand description
AgenticEvolve — pattern library engine for AI agents.
Crystallizes verified code patterns for reuse, providing ~80% of function bodies from patterns on subsequent builds.
Re-exports§
pub use types::error::EvolveError;pub use types::error::EvolveResult;pub use types::ids::EvolveId;pub use types::ids::PatternId;pub use types::ids::SkillId;pub use types::match_result::MatchContext;pub use types::match_result::MatchResult;pub use types::match_result::MatchScore;pub use types::pattern::FunctionSignature;pub use types::pattern::Language;pub use types::pattern::Pattern;pub use types::pattern::PatternVariable;pub use types::skill::CrystallizedSkill;pub use types::skill::SkillMetadata;pub use types::skill::SuccessfulExecution;
Modules§
- bridges
- Sister integration bridge traits for AgenticEvolve.
- cache
- Cache layer — LRU cache with TTL, invalidation, and metrics.
- collective
- Collective learning — track usage, success, decay, and promotion.
- composition
- Composition — combine multiple patterns and fill gaps.
- crystallization
- Crystallization — extract patterns from successful code.
- matching
- Pattern matching — find the best pattern for a given function signature.
- metrics
- Metrics layer — token tracking, audit logging, conservation scoring.
- optimization
- Pattern optimization — storage optimization and cache management.
- query
- Query layer — intent-based extraction, delta queries, token budgets, pagination.
- storage
- Pattern storage — save, load, index, and version patterns.
- types
- Core types for AgenticEvolve.