MCP Server

Dakera ships as a native MCP server — 83 tools for Claude Desktop, Claude Code, Cursor, Windsurf, and any MCP-compatible client. No code changes required.

What is MCP? The Model Context Protocol allows AI assistants to call external tools. Dakera exposes its complete API as 83 callable MCP tools.

Install the MCP server

Two installation methods are available today. Choose whichever fits your workflow:

cargo install dakera-mcp       # Rust 1.70+
dakera-mcp --version

Don't have Rust? curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

docker run --rm -i \
  -e DAKERA_API_URL=http://host.docker.internal:3300 \
  -e DAKERA_API_KEY=your-api-key \
  ghcr.io/dakera-ai/dakera-mcp:latest

Point the MCP server at your Dakera instance

Set two environment variables — DAKERA_API_URL is the address of your running Dakera server (the IP or hostname where you deployed it), and DAKERA_API_KEY is the root key you set when starting the server.

Where is my server? If you started Dakera with docker run -p 3300:3300 ... on a remote host, use http://<SERVER_IP>:3300. For a local dev machine use http://localhost:3300. For Kubernetes with an ingress, use your ingress hostname.

# Quick test — verify connectivity before wiring up any client:
export DAKERA_API_URL=http://<YOUR_SERVER_IP>:3300
export DAKERA_API_KEY=<your-api-key>
dakera-mcp --test-connection
# Connected to Dakera v0.11.55 — 83 tools available

Connect your AI client

Replace <YOUR_SERVER_IP> with your Dakera server address and <your-api-key> with your root API key.

// Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json
// Claude Code: .mcp.json (project) or ~/.claude/settings.json (global)
// Cursor: ~/.cursor/mcp.json
// Windsurf: ~/.codeium/windsurf/mcp_config.json
{
  "mcpServers": {
    "dakera": {
      "command": "/usr/local/bin/dakera-mcp",
      "env": {
        "DAKERA_API_URL": "http://<YOUR_SERVER_IP>:3300",
        "DAKERA_API_KEY": "<your-api-key>"
      }
    }
  }
}

Restart your client after editing. For Zed, use context_servers with a command.path key instead of mcpServers.

Verify the connection

dakera-mcp --test-connection  # or ask Claude: "List the Dakera MCP tools available"

Tool categories (83 tools)

Category	Key tools	Count
Memory	store, recall, batch_recall, search, memory_get, memory_update, memory_importance, forget, consolidate, batch_forget	10
Memory Feedback	memory_feedback, memory_feedback_get, agent_feedback_summary	3
Sessions	session_start, session_end, session_list, session_get, session_memories	5
Agents	agent_stats, agent_memories, agent_sessions	3
Knowledge	knowledge_graph, knowledge_summarize, knowledge_deduplicate, knowledge_network_cross_agent	4
Knowledge Graph	graph_link_memory, graph_traverse, graph_path, graph_export, kg_query, kg_traverse, kg_export	7
Entity Extraction	extract_entities, auto_tag, memory_entities, extract, extractor_get, extractor_set, entity_types_get, entity_types_set	8
Namespaces	namespace_list, namespace_get, namespace_create, namespace_delete, namespace_configure, namespace_key_create, namespace_key_list, namespace_key_delete, namespace_key_usage	9
Vectors	vector_upsert, vector_upsert_columns, vector_query, vector_batch_query, vector_multi_search, vector_unified_query, vector_delete, vector_bulk_update, vector_bulk_delete, vector_count, vector_export, vector_aggregate, vector_explain, vector_warm	14
Full-text & Inference	fulltext_index, fulltext_search, fulltext_stats, fulltext_delete, upsert_text, text_query, batch_query_text, hybrid_search	8
Decay & AutoPilot	decay_config_get, decay_config_set, decay_stats, autopilot_trigger, autopilot_status	5
Admin & Other	audit_query, memory_export, memory_import, encryption_rotate_key, agent_feedback_summary, memory_policy_get, memory_policy_set	7

Complete MCP tool reference

Every tool listed below is available as a callable MCP action. All tools accept JSON parameters and return JSON responses. Prefix each tool name with dakera_ when calling from your AI client.

Memory operations (11 tools)

Tool	Description	Key parameters
`store`	Store a memory for an agent	`agent_id` req, `content` req, `importance` opt, `tags` opt, `memory_type` opt, `session_id` opt, `expires_at` opt
`recall`	Semantic recall — returns most relevant memories for a query	`agent_id` req, `query` req, `top_k` opt, `min_importance` opt, `include_associated` opt, `since` opt, `until` opt
`recall_associated`	Recall with deep knowledge-graph context enrichment (1–3 hops)	`agent_id` req, `query` req, `associated_memories_depth` opt, `associated_memories_min_weight` opt
`batch_recall`	Filter-based bulk recall without semantic search	`agent_id` req, `tags` opt, `min_importance` opt, `max_importance` opt, `created_after` opt, `created_before` opt, `memory_type` opt
`forget`	Delete memories by ID or tag filter	`agent_id` req, `memory_ids` opt, `tags` opt
`batch_forget`	Bulk delete all memories matching filter predicates	`agent_id` req, `tags` opt, `min_importance` opt, `memory_type` opt
`search`	Search with tag filtering and importance thresholds	`agent_id` req, `query` req, `top_k` opt, `tags` opt, `memory_type` opt
`memory_get`	Get a specific memory by ID	`agent_id` req, `memory_id` req
`memory_update`	Update content, importance, or tags (re-embeds if content changes)	`agent_id` req, `memory_id` req, `content` opt, `importance` opt, `tags` opt
`memory_importance`	Batch-update importance scores for multiple memories	`agent_id` req, `updates` req (array of {memory_id, importance})
`consolidate`	Merge related memories into a synthesized summary	`agent_id` req, `memory_ids` req

Memory feedback (3 tools)

Tool	Description	Key parameters
`memory_feedback`	Submit feedback signal — upvotes increase importance, downvotes decrease	`memory_id` req, `signal` req (upvote \| downvote \| flag)
`memory_feedback_get`	Retrieve feedback history for a memory	`memory_id` req
`agent_feedback_summary`	Aggregated feedback stats for all of an agent's memories	`agent_id` req

Sessions (5 tools)

Tool	Description	Key parameters
`session_start`	Start a new session to group related memories	`agent_id` req, `metadata` opt
`session_end`	End an active session with optional summary	`session_id` req, `summary` opt
`session_list`	List sessions for an agent	`agent_id` req, `active_only` opt
`session_get`	Get session details including metadata and summary	`session_id` req
`session_memories`	List all memories associated with a session	`session_id` req

Agents (3 tools)

Tool	Description	Key parameters
`agent_stats`	Memory count, session count, storage usage, and top tags	`agent_id` req
`agent_memories`	List all memories for an agent, ordered by recency	`agent_id` req, `limit` opt, `offset` opt
`agent_sessions`	List all sessions (active and completed) for an agent	`agent_id` req

Knowledge management (4 tools)

Tool	Description	Key parameters
`knowledge_graph`	Build a knowledge graph from a seed memory via embedding similarity	`agent_id` req, `memory_id` req, `depth` opt, `min_similarity` opt
`knowledge_summarize`	Summarize a set of memories into a consolidated memory	`agent_id` req, `memory_ids` req, `target_type` opt
`knowledge_deduplicate`	Find and optionally merge duplicate memories	`agent_id` req, `threshold` opt, `dry_run` opt
`knowledge_network_cross_agent`	Build a cross-agent memory network spanning all agents	`agent_ids` opt, `min_similarity` opt, `max_nodes_per_agent` opt, `max_cross_edges` opt

Knowledge graph traversal (7 tools)

Tool	Description	Key parameters
`graph_traverse`	BFS traversal from a starting memory in the knowledge graph	`memory_id` req, `depth` opt
`graph_path`	Find shortest path between two memories	`from_id` req, `to_id` req
`graph_link_memory`	Create an explicit edge between two memories	`memory_id` req, `target_id` req
`graph_export`	Export full knowledge graph for an agent as edge list	`agent_id` req
`kg_traverse`	Filtered graph traversal with edge-type and weight filters	`agent_id` req, `root_id` req, `max_depth` opt, `edge_type` opt, `min_weight` opt
`kg_query`	Query the knowledge graph using a filter DSL	`agent_id` req, `edge_type` opt, `min_weight` opt, `limit` opt
`kg_export`	Export full knowledge graph as JSON or GraphML	`agent_id` req, `format` opt (json \| graphml)

Entity extraction (8 tools)

Tool	Description	Key parameters
`auto_tag`	Extract entities using GLiNER zero-shot NER + rule-based pre-pass	`content` req, `entity_types` opt
`extract`	Extract structured info (entities, topics, key phrases, summary) from text	`text` req, `namespace` opt, `entity_types` opt, `extractor_override` opt
`extract_entities`	Named entity extraction via the dakera-ode GLiNER pipeline	`content` req, `agent_id` req, `memory_id` opt, `entity_types` opt
`memory_entities`	Get entities extracted when a memory was stored	`memory_id` req
`entity_types_get`	Get entity extraction config for a namespace	`namespace` req
`entity_types_set`	Configure entity extraction for a namespace	`namespace` req, `extract_entities` req, `entity_types` req
`extractor_get`	Read default extraction provider for a namespace	`namespace` req
`extractor_set`	Set extraction provider (OpenAI, Anthropic, Ollama, ONNX)	`namespace` req, `provider` req, `model` opt, `base_url` opt

Namespace management (9 tools)

Tool	Description	Key parameters
`namespace_list`	List all namespaces	(none)
`namespace_get`	Get namespace details (vector count, dimensions, index stats)	`namespace` req
`namespace_create`	Create a namespace with specified dimensions and distance metric	`name` req, `dimension` req, `distance` opt
`namespace_delete`	Delete a namespace and all its vectors (irreversible)	`namespace` req
`namespace_configure`	Create-or-update namespace configuration	`namespace` req, `dimension` req, `distance` opt
`namespace_key_create`	Create a namespace-scoped API key	`namespace` req, `name` req, `scope` req, `expires_in_days` opt
`namespace_key_list`	List all API keys for a namespace	`namespace` req
`namespace_key_delete`	Revoke a namespace-scoped API key	`namespace` req, `key_id` req
`namespace_key_usage`	Get usage statistics for a namespace key	`namespace` req, `key_id` req

Vector operations (14 tools)

Tool	Description	Key parameters
`vector_upsert`	Insert or update vectors with optional metadata	`namespace` req, `vectors` req
`vector_upsert_columns`	Column-format upsert — more efficient for batch operations	`namespace` req, `ids` req, `vectors` req, `attributes` opt
`vector_query`	Nearest-neighbor similarity search	`namespace` req, `vector` req, `top_k` opt, `filter` opt
`vector_batch_query`	Multiple similarity searches in parallel	`namespace` req, `queries` req
`vector_multi_search`	Multi-vector search with positive/negative vectors, MMR diversity	`namespace` req, `positive_vectors` req, `negative_vectors` opt, `enable_mmr` opt, `mmr_lambda` opt
`vector_unified_query`	Flexible ranking — vector, full-text, or attribute ordering	`namespace` req, `rank_by` req, `top_k` opt, `filter` opt
`vector_delete`	Delete vectors by ID	`namespace` req, `ids` req
`vector_bulk_update`	Update metadata on all vectors matching a filter	`namespace` req, `filter` req, `update` req
`vector_bulk_delete`	Delete all vectors matching a filter	`namespace` req, `filter` req
`vector_count`	Count vectors, optionally filtered	`namespace` req, `filter` opt
`vector_export`	Export vectors with pagination	`namespace` req, `top_k` opt, `cursor` opt, `include_vectors` opt
`vector_aggregate`	Compute aggregations (Count, Sum, Avg, Min, Max)	`namespace` req, `aggregate_by` req, `group_by` opt, `filter` opt
`vector_explain`	Explain query execution plan — index selection, cost estimates	`namespace` req, `query_type` req, `execute` opt, `verbose` opt
`vector_warm`	Pre-load vectors into memory for faster queries	`namespace` req, `vector_ids` opt

Full-text, text inference & hybrid search (8 tools)

Tool	Description	Key parameters
`fulltext_index`	Index documents for BM25 full-text search	`namespace` req, `documents` req
`fulltext_search`	BM25 keyword search	`namespace` req, `query` req, `top_k` opt, `filter` opt
`fulltext_delete`	Remove documents from the full-text index	`namespace` req, `ids` req
`fulltext_stats`	Index statistics (document count, term count)	`namespace` req
`upsert_text`	Upsert text documents with automatic server-side embedding	`namespace` req, `documents` req, `model` opt
`text_query`	Query using natural language — text is embedded server-side	`namespace` req, `text` req, `top_k` opt, `filter` opt
`batch_query_text`	Batch text queries with automatic embedding	`namespace` req, `queries` req, `top_k` opt
`hybrid_search`	Combined vector similarity + BM25 full-text search	`namespace` req, `text` req, `vector` opt, `top_k` opt, `vector_weight` opt

Decay & AutoPilot (5 tools)

Tool	Description	Key parameters
`decay_config_get`	Get current decay settings (strategy, half-life, min importance)	(none)
`decay_config_set`	Update decay settings at runtime (no restart)	`strategy` opt, `half_life_hours` opt, `min_importance` opt
`decay_stats`	Get cumulative decay statistics and last-cycle details	(none)
`autopilot_status`	AutoPilot config and last-run stats (deduped, consolidated)	(none)
`autopilot_trigger`	Force an immediate AutoPilot cycle	`action` req (dedup \| consolidate \| all)

Memory lifecycle policy (2 tools)

Tool	Description	Key parameters
`memory_policy_get`	Get lifecycle policy — TTLs, decay curves, spaced repetition, consolidation, rate limits	`namespace` req
`memory_policy_set`	Update lifecycle policy per namespace	`namespace` req, per-type TTLs and decay curves, `consolidation_enabled` opt, `rate_limit_enabled` opt

Audit & import/export (4 tools)

Tool	Description	Key parameters
`audit_query`	Query business-event audit log (stored, recalled, forgotten, etc.)	`agent_id` opt, `event_type` opt, `from` opt, `to` opt, `limit` opt
`memory_export`	Export all memories in a portable format	`agent_id` req, `format` opt (jsonl \| csv \| mem0 \| zep)
`memory_import`	Import memories from JSONL, CSV, Mem0 JSON, or Zep JSON	`agent_id` req, `data` req, `format` opt
`encryption_rotate_key`	Zero-downtime AES-256-GCM encryption key rotation	`new_key` req, `namespace` opt