{
  "markdown": "# ⚡ dakera-mcp\n\n[![CI](https://github.com/Dakera-AI/dakera-mcp/actions/workflows/ci.yml/badge.svg)](https://github.com/Dakera-AI/dakera-mcp/actions/workflows/ci.yml) [![Crate](https://img.shields.io/crates/v/dakera-mcp?logo=rust)](https://crates.io/crates/dakera-mcp) [![npm](https://img.shields.io/npm/v/%40dakera-ai%2Fdakera-mcp?logo=npm)](https://www.npmjs.com/package/@dakera-ai/dakera-mcp) [![Downloads](https://img.shields.io/crates/d/dakera-mcp)](https://crates.io/crates/dakera-mcp) [![License: MIT](https://img.shields.io/github/license/Dakera-AI/dakera-mcp)](LICENSE) [![LoCoMo 88.2%](https://img.shields.io/badge/LoCoMo-88.2%25-22c55e?style=flat-square)](https://dakera.ai/benchmark) [![Glama](https://glama.ai/mcp/servers/Dakera-AI/dakera-mcp/badge)](https://glama.ai/mcp/servers/Dakera-AI/dakera-mcp) [![Docs](https://img.shields.io/badge/docs-dakera.ai%2Fdocs-3b82f6?style=flat-square)](https://dakera.ai/docs) [![dakera.ai](https://img.shields.io/badge/dakera.ai-website-22c55e?style=flat-square)](https://dakera.ai) [![Playground](https://img.shields.io/badge/playground-try%20it-ff6b35?style=flat-square)](https://dakera.ai/playground)\n\nMCP server for Dakera AI. Gives any MCP-compatible AI agent persistent, queryable memory — with smart token management built in.\n\nWorks with Claude, Claude Code, and any MCP-compatible framework.\n\nPart of [Dakera AI](https://dakera.ai) — the memory engine for AI agents.\n\n> The Dakera memory engine scores **88.2% Recall@20 on LoCoMo** (1,540 questions · LLM-judge scored) — [benchmark details](https://dakera.ai/benchmark)\n\n---\n\n## Architecture: 14 core tools + on-demand discovery\n\nStarting every agent session with 60+ tool schemas wastes ~15K tokens before you write a single message. dakera-mcp solves this with **hybrid tool exposure**:\n\n- **14 tools loaded by default** — the 12 highest-frequency memory operations + 2 meta-discovery tools\n- **On-demand expansion** — use `dakera_discover_tools` and `dakera_load_tools` to fetch additional tool schemas only when you need them\n\n### Default tool set (core profile)\n\n| Tool | Purpose |\n|---|---|\n| `dakera_store` | Store a memory with importance, tags, and type |\n| `dakera_recall` | Semantic recall by query text |\n| `dakera_search` | Advanced memory search with tag/type filters |\n| `dakera_session_start` | Start a session to group related memories |\n| `dakera_session_end` | End a session with optional summary |\n| `dakera_batch_recall` | Bulk filter-based recall (by tags, importance, time) |\n| `dakera_forget` | Delete specific memories by ID |\n| `dakera_hybrid_search` | Combined vector + BM25 search |\n| `dakera_fulltext_search` | BM25 full-text search |\n| `dakera_knowledge_graph` | Build a knowledge graph from a seed memory |\n| `dakera_extract` | Extract entities and structure from free-form text |\n| `dakera_batch_forget` | Bulk delete by tags, type, or time range |\n| `dakera_discover_tools` | Search the full tool catalog by keyword or tier |\n| `dakera_load_tools` | Load full schemas for specific tools on demand |\n\n### Profiles & token cost\n\n| Profile | Tools | ~Tokens | How to enable |\n|---|---|---|---|\n| **core** | 14 | ~2,964 | Default — always loaded |\n| **admin** | 32 | ~5,975 | `DAKERA_MCP_PROFILE=admin` |\n| **power** | 69 | ~13,205 | `DAKERA_MCP_PROFILE=power` |\n| **all** | 87 | ~16,212 | `DAKERA_MCP_PROFILE=all` |\n\n### Accessing additional tools\n\n```\n# In your agent: discover what's available\ndakera_discover_tools(tier=\"power\")\n→ returns names + descriptions, no schemas loaded\n\n# Load schemas for the tools you want\ndakera_load_tools(tools=[\"dakera_consolidate\", \"dakera_agent_stats\"])\n→ returns full inputSchema for each tool\n```\n\n### Profile selection\n\nThe profile controls which tools appear in `tools/list`. Three ways to set it:\n\n**1. Per-request** (in `tools/list` params):\n```json\n{\"profile\": \"power\"}\n```\n\n**2. Environment variable** (applies to all requests):\n```bash\nDAKERA_MCP_PROFILE=power\n```\n\n**3. Default**: `core` (14 tools, ~2,964 tokens)\n\n---\n\n## Run Dakera\n\nThe MCP server connects to a Dakera memory server. You need one running first:\n\n```bash\ndocker run -d \\\n  --name dakera \\\n  -p 3300:3000 \\\n  -e DAKERA_ROOT_API_KEY=dk-mykey \\\n  ghcr.io/dakera-ai/dakera:latest\n```\n\nFor persistent storage (recommended):\n\n```bash\ncurl -sSfL https://raw.githubusercontent.com/Dakera-AI/dakera-deploy/main/docker-compose.yml \\\n  -o docker-compose.yml\nDAKERA_API_KEY=dk-mykey docker compose up -d\n\ncurl http://localhost:3000/health  # → {\"status\":\"ok\"}\n```\n\nFull deployment guide (Docker Compose, Kubernetes, Helm): [dakera-deploy](https://github.com/Dakera-AI/dakera-deploy)\n\n---\n\n## Install\n\n### npm / npx (Node.js 18+)\n\n```bash\n# Global install\nnpm install -g @dakera-ai/dakera-mcp\n\n# Or run directly without installing\nnpx @dakera-ai/dakera-mcp\n```\n\n### Homebrew (macOS / Linux)\n\n```bash\nbrew install dakera-ai/tap/dakera-mcp\n```\n\n### Cargo\n\n```bash\ncargo install dakera-mcp\n```\n\n### Docker\n\n```bash\ndocker pull ghcr.io/dakera-ai/dakera-mcp:latest\n```\n\n### Binary download\n\nPre-built binaries for macOS, Linux, and Windows are available on the [releases page](https://github.com/Dakera-AI/dakera-mcp/releases).\n\n| Platform | File |\n|---|---|\n| macOS (Apple Silicon) | `dakera-mcp-aarch64-apple-darwin.tar.gz` |\n| macOS (Intel) | `dakera-mcp-x86_64-apple-darwin.tar.gz` |\n| Linux x64 | `dakera-mcp-x86_64-unknown-linux-musl.tar.gz` |\n| Linux arm64 | `dakera-mcp-aarch64-unknown-linux-musl.tar.gz` |\n| Windows x64 | `dakera-mcp-x86_64-pc-windows-msvc.zip` |\n\n---\n\n## Connect\n\nAdd to `.mcp.json` (Claude Code) or `claude_desktop_config.json` (Claude Desktop):\n\n```json\n{\n  \"mcpServers\": {\n    \"dakera\": {\n      \"command\": \"dakera-mcp\",\n      \"env\": {\n        \"DAKERA_API_URL\": \"http://localhost:3300\",\n        \"DAKERA_API_KEY\": \"your-key\"\n      }\n    }\n  }\n}\n```\n\nTo start with the power profile (exposes 68 tools):\n\n```json\n{\n  \"mcpServers\": {\n    \"dakera\": {\n      \"command\": \"dakera-mcp\",\n      \"env\": {\n        \"DAKERA_API_URL\": \"http://localhost:3300\",\n        \"DAKERA_API_KEY\": \"your-key\",\n        \"DAKERA_MCP_PROFILE\": \"power\"\n      }\n    }\n  }\n}\n```\n\n## Why This Exists\n\nAI agents forget everything when the session ends. Dakera fixes that. This MCP server gives your agent a persistent memory layer with zero infrastructure overhead — point it at a Dakera instance and it works.\n\nThe 14-tool default keeps your context window lean. The meta-tools let you expand on demand when you need advanced operations like bulk vector upsert, knowledge graph traversal, or memory federation.\n\n→ [dakera.ai](https://dakera.ai) for hosted instance  \n→ Self-host with [dakera-deploy](https://github.com/dakera-ai/dakera-deploy)\n\n## Documentation\n\n→ [Full docs](https://dakera.ai/docs)  \n→ [MCP reference](https://dakera.ai/docs/mcp)\n\n## Related\n\n| Repo | What it is |\n|---|---|\n| [dakera-py](https://github.com/dakera-ai/dakera-py) | Python SDK |\n| [dakera-js](https://github.com/dakera-ai/dakera-js) | TypeScript SDK |\n| [dakera-cli](https://github.com/dakera-ai/dakera-cli) | CLI |\n| [dakera-deploy](https://github.com/dakera-ai/dakera-deploy) | Self-host Dakera |\n\n---\n\n**[dakera.ai](https://dakera.ai)** · [Documentation](https://dakera.ai/docs) · [Request Early Access](https://dakera.ai#cta)\n\n<sub>Part of the Dakera AI open-core ecosystem. Built with Rust. Self-hosted. Zero dependencies.</sub>\n",
  "bytes": 7345,
  "sha": "6d485f8a84a0b6f6c5764a9dc0be9fb4bad4a71c5c4dfcf4f4d3864903e1ecd3",
  "repo_slug": "dakera-ai/dakera-mcp",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_github_dakera_ai_dakera_mcp_f34500b1/readme"
}