MCP 服务器

Adaptive Recall

Adaptive MCP memory system for AI applications. Learns which retrieval strategies work for your data, scores results using cognitive science models, builds a knowledge graph automatically, and validates every parameter change against real query history before adopting it. Patent pending.

README

Adaptive Recall

Adaptive memory system for AI applications. Patent pending.

adaptiverecall.com | Documentation | Sign Up Free

What It Does

Adaptive Recall is a hosted memory server that stores, retrieves, and manages long-term memory for AI applications. It connects via MCP or REST API.

Multi-strategy retrieval: four search strategies run in parallel (vector similarity, temporal recency, full-text keyword, knowledge graph traversal) and the system learns which to prioritize for each type of query
Cognitive scoring: results ranked using ACT-R activation modeling from cognitive science, factoring in recency, access frequency, entity connections, and validated confidence
Knowledge graph: entities and relationships extracted automatically from stored memories, used as a retrieval pathway alongside text similarity
Memory lifecycle: memories progress through stages, gain or lose confidence based on corroborating evidence, and fade naturally when unused
Self-improving: ML models train on your usage patterns, every parameter change must pass statistical validation against real query history before being adopted
Retrieval quality monitoring: the system verifies its own retrieval consistency and identifies knowledge gaps

Connect

MCP Configuration

Add to your MCP client config (Claude Code, Codex, Cursor, or any MCP-compatible tool):

{
  "mcpServers": {
    "adaptive-recall": {
      "type": "url",
      "url": "https://YOUR_SERVER_URL/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

For Claude Code, add this to .mcp.json in your project or ~/.claude/settings.json for global access. For Gemini CLI, add to ~/.gemini/settings.json using httpUrl instead of url. For Codex, add to your Codex MCP configuration.

REST API

Every action is also available as an HTTP endpoint at https://YOUR_SERVER_URL/v1/. All requests require a Bearer token in the Authorization header.

Actions

Action	Description
store	Save a new memory. Generates embeddings and extracts entities automatically.
recall	Search memories using multi-strategy retrieval with cognitive scoring.
update	Modify an existing memory. Re-embeds automatically if content changes.
forget	Remove a memory by ID or by finding the closest match to a query.
graph	Explore the knowledge graph, traversing entity relationships by name and depth.
status	System health, memory counts, confidence distribution, and knowledge gap detection.
snapshot	Get a formatted overview of stored memories, organized by type.
feedback	Send feedback directly to the Adaptive Recall developers.