All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
- 17 MCP tools for persistent cross-session memory
- PostgreSQL + pgvector storage backend with full-text search (tsvector/GIN)
- Hybrid search combining FTS BM25 (50%) and vector similarity (50%)
- Semantic search via fastembed-rs (BGE-M3, 1024d, 100+ languages)
- Infinite Memory with hierarchical summaries and drill-down API
- Privacy filtering with
<private>tag support across all storage paths - Context-aware deduplication (CREATE/UPDATE/SKIP) to prevent observation drift
- 65 HTTP API endpoints with SSE real-time updates
- Web viewer with dark theme UI
- CLI with hooks for OpenCode integration (context, session-init, observe, summarize)
- Pending queue with crash recovery, visibility timeout, and dead letter queue
- Project exclusion filtering via
OPENCODE_MEM_EXCLUDED_PROJECTS - Configurable low-value observation filter (SPOT pattern)
- Knowledge base with global knowledge entries (skill, pattern, gotcha, architecture, tool_usage)
- Save memory tool for direct observation storage bypassing LLM compression