Dakera Mcp
Give Claude Code or Cursor a persistent, decay-weighted memory layer so agents remember context across sessions without stuffing the context window.
Overview
Dakera MCP is a MCP server for the Build phase that provides decay-weighted vector memory and 83 tools for store, recall, search, and knowledge graphs for AI agents.
What is this MCP server?
- 83 MCP tools for store, recall, semantic search, and knowledge-graph operations
- Decay-weighted vector memory so older facts fade unless reinforced
- stdio transport via OCI image ghcr.io/dakera-ai/dakera-mcp
- Local-first style memory suitable for long-running agent workflows
- Version 0.9.8 in the official MCP server schema
- 83 MCP tools documented in the server description
- Server version 0.9.8
- stdio transport via OCI identifier ghcr.io/dakera-ai/dakera-mcp:0.9.8
Community signal: 3 GitHub stars.
What problem does it solve?
Coding agents forget everything when the session ends, so solo builders either burn tokens repeating context or ship agents that feel amnesiac.
Who is it for?
Indie builders adding durable memory to Claude Code or Cursor workflows, RAG-style agents, or multi-step coding pipelines that need recall without a custom vector DB integration.
Skip if: Teams that only need a simple notes search in Obsidian or read-only analytics on an existing SaaS backend with no agent memory requirements.
What do I get? / Deliverables
After you register Dakera MCP, your agent can persist and retrieve weighted memories and graph-linked knowledge through standard MCP tool calls across sessions.
- MCP-configured agent with memory store/recall/search tools
- Decay-weighted retention behavior for agent-held facts
- Knowledge-graph-oriented tool calls usable from the agent loop
Recommended MCP Servers
Journey fit
Agent memory is core product infrastructure you wire up while building AI-assisted apps and custom agents, not a one-off launch task. Dakera MCP is purpose-built agent tooling: vector store, recall, search, and knowledge graphs exposed as MCP tools for coding agents.
How it compares
Long-horizon agent memory MCP server, not a single-purpose research or analytics connector.
Common Questions / FAQ
Who is Dakera MCP for?
Solo and indie developers building AI agents or heavy MCP-driven coding workflows who want vector memory and knowledge graphs without wiring a separate memory stack by hand.
When should I use Dakera MCP?
Use it during Build when you are integrating agent persistence, or when Operate/iterate cycles need the agent to remember prior decisions, bugs, and customer context with decay-weighted recall.
How do I add Dakera MCP to my agent?
Add the stdio MCP entry pointing at the OCI image ghcr.io/dakera-ai/dakera-mcp:0.9.8 in your Claude Code, Cursor, or Windsurf MCP config per your client’s server.json format.