Alaya Mcp
Alaya is a neuroscience-inspired memory engine that gives AI agents local knowledge graphs with forgetting curves. It exposes 13 MCP tools including remember, recall, learn, and maintain. The engine stores conversations as episodes, consolidates them into semantic knowledge nodes, and applies Bjork dual-strength forgetting so stale memories decay while frequently retrieved ones strengthen. Categories emerge automatically from usage patterns. Everything lives in a single SQLite file with no external services required. You can run it via npx or build from Rust source. The recall tool uses hybrid retrieval with spreading activation across the knowledge graph, returning only relevant memories instead of dumping full context. Configure optional LLM credentials for auto-consolidation or let it prompt the agent to call learn manually.
- GitHub stars
- 12
- tools
- 13
- auth
- Secret needed
- transport
- STDIO
- env vars
- 4
- last updated
- Mar 27, 2026
Tools
13 toolsPublic tool metadata for what this MCP can expose to an agent.
rememberStore a conversation message (auto-prompts consolidation after 10 episodes)
Store a conversation message (auto-prompts consolidation after 10 episodes)
No parameter schema in public metadata yet.
recallSearch memory with hybrid retrieval (+ category boost)
Search memory with hybrid retrieval (+ category boost)
No parameter schema in public metadata yet.
learnTeach extracted knowledge directly — agent extracts facts and calls this
Teach extracted knowledge directly — agent extracts facts and calls this
No parameter schema in public metadata yet.
statusRich memory statistics: episodes, knowledge breakdown, categories, graph, embeddings
Rich memory statistics: episodes, knowledge breakdown, categories, graph, embeddings
No parameter schema in public metadata yet.
preferencesGet learned user preferences
Get learned user preferences
No parameter schema in public metadata yet.
knowledgeGet distilled semantic facts (+ category filter)
Get distilled semantic facts (+ category filter)
No parameter schema in public metadata yet.
maintainRun memory cleanup (dedup, decay)
Run memory cleanup (dedup, decay)
No parameter schema in public metadata yet.
purgeDelete memories by session, age, or all
Delete memories by session, age, or all
No parameter schema in public metadata yet.
categoriesList emergent categories with stability filter
List emergent categories with stability filter
No parameter schema in public metadata yet.
neighborsGraph neighbors via spreading activation
Graph neighbors via spreading activation
No parameter schema in public metadata yet.
node_categoryWhich category a node belongs to
Which category a node belongs to
No parameter schema in public metadata yet.
import_claude_memImport observations from a claude-mem database
Import observations from a claude-mem database
No parameter schema in public metadata yet.
import_claude_codeImport conversation history from Claude Code JSONL files
Import conversation history from Claude Code JSONL files
No parameter schema in public metadata yet.
Sponsors
Agent CaffeineSponsoredClose your lid, come back to finished work
SnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthly
WaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthly
knobsSponsoredSee your site on any phone, any day, in any languageConfiguration
ALAYA_LLM_API_KEYsecretAPI key for auto-consolidation (enables LLM-powered knowledge extraction)
ALAYA_LLM_API_URLOpenAI-compatible chat completions endpoint
ALAYA_LLM_MODELModel name (e.g. gpt-4o-mini, haiku, gemini-flash)
ALAYA_DBPath to SQLite database
Readme
Details
- Repository
- github.com/securityronin/alaya
- Categories
- AI & LLM Tools, Documents & Knowledge
- Package
alaya-mcp- Transport
- STDIO
- Auth
- Required
- Registry
- MCP registry, active
Related AI & LLM Tools MCP servers
View all
ChromaDB Remote MCP Servermeloncafe/chromadb-remote-mcpRemote ChromaDB vector database MCP server with streamable HTTP transport12stars
Pindocvar-gg/pindocCode-pinned team memory for AI coding agents — typed artifacts, MCP-native, self-host.12stars
Velociraghaseebkhalid1507/velociragLightning-fast RAG for AI agents. 4-layer fusion, ONNX Runtime, sub-200ms search.10stars
Octen Mcpocten-team/octen-mcpURL to LLM-ready markdown — plus per-page category, page_structure, and query-driven highlights.10stars
Decomposeechology-io/decomposeDecompose text into classified semantic units. Authority, risk, attention. No LLM.9stars