data-ai MCP Server
Xmem is a India's First multi-modal, multi-agentic long‑term memory layer for AI agents.
Discovered via github-topic:mcp and last synced 4mo ago.
Install instructions not detected yet
Check the source repository for the latest setup steps.
XMem (Gemini 3-flash)
Default
`60`
62.23
—
74.11
`openrouter,gemini,claude,openai`
Used For
`gemini-embedding-001`
Single-Hop (%)
`gemini-2.5-flash-lite`
`0.4`
What It Stores
—
### Agentic Ingestion Pipeline Every conversation turn flows through a **7-stage LangGraph pipeline**: ``` Input → Classify → Extract (parallel) → Judge → Weave → Store ``` 1. **Classifier** routes input to the relevant domains 2. **Domain Agents** (Profiler, Temporal, Summarizer, Code, Snippet, Image) extract structured data in parallel 3. **Judge Agent** compares each extraction against existing memory and decides: `ADD`, `UPDATE`, `DELETE`, or `NOOP` 4. **Weaver** deterministically executes the Judge's decisions across all storage backends This means XMem doesn't just append — it **maintains** memory. Tell it you switched from Python to Go? The Judge updates your profile. Mention a meeting got rescheduled? The temporal record is corrected, not duplicated. ### Two-Step Agentic Retrieval When you query XMem, retrieval is not a simple vector search. The LLM itself decides *what* to look up: 1. **Tool Selection**: The retrieval LLM analyzes your query and calls the appropriate search tools — `SearchProfile`, `SearchTemporal`, `SearchSummary`, `SearchSnippet` — potentially multiple in parallel 2. **Synthesis**: Results from all search tools are aggregated and the LLM generates a cited answer with source references This means asking *"What's my preferred tech stack and when did I last refactor the auth module?"* triggers both a profile lookup and a temporal search — automatically. ### Code Scanner (XIDE) XMem can index entire Git repositories and build a queryable knowledge graph of your codebase: - **AST Parsing**: Deterministic parsing (no LLM needed) for Python, TypeScript, and JavaScript. Extracts functions, classes, methods, imports, and call graphs. - **Incremental Scanning**: Uses `git diff` to only re-process changed files - **Knowledge Graph**: Builds a Neo4j graph with `IMPORTS`, `CALLS`, and `ANNOTATES` relationships between symbols - **Chat With Your Code**: Stream-based chat interface that retrieves relevant code context from your indexed repos ### Multi-LLM Orchestration with Fallback XMem isn't locked to one provider. It orchestrates across **Gemini, Claude, OpenAI, OpenRouter, and Amazon Bedrock** with automatic failover: ``` gemini → claude → openai → bedrock ``` If your primary LLM API rate-limits or goes down, XMem silently falls back to the next provider. Your memory pipeline never breaks. Each agent can even be pinned to a specific model — use Gemini for classification but Claude for retrieval synthesis. ### Multi-Storage Backend Each memory domain maps to the storage engine best suited for it:
`768`
63.79
—