AgentRecall
AI Session Memory with Think-Execute-Reflect Quality Loops — give your agent a brain that survives every session. Built on the Intelligent Distance principle.
What is AgentRecall?
AgentRecall is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to ai session memory with think-execute-reflect quality loops — give your agent a brain that survives every session. built on the intelligent distance principle.
AI Session Memory with Think-Execute-Reflect Quality Loops — give your agent a brain that survives every session. Built on the Intelligent Distance principle.
This server falls under the Knowledge & Memory category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- AI Session Memory with Think-Execute-Reflect Quality Loops —
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx agentrecall-mcpConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use AgentRecall
AgentRecall MCP is a persistent session memory system for AI agents that implements a Think-Execute-Reflect quality loop built on the Intelligent Distance principle. It gives agents a structured memory store that survives across sessions, organized into semantic rooms (facts with wikilinks), procedural skills (IF-THEN rules), episodic journals (session entries), behavioral corrections (rules with KPIs), and cross-project awareness — all stored locally as markdown files with no cloud dependency or API keys required. Developers use AgentRecall to build Claude Code, Cursor, and VS Code Copilot workflows where the agent accumulates institutional knowledge, recalls relevant context, and improves its behavior over time.
Prerequisites
- Node.js 18 or later installed
- An MCP-compatible client: Claude Code, Cursor, VS Code with Copilot, or Windsurf
- No API keys required — all storage is local markdown files in ~/.agent-recall/
Add AgentRecall to Claude Code
Register the AgentRecall MCP server with Claude Code using the `claude mcp add` command. The server runs via npx without any global installation.
claude mcp add --scope user agent-recall -- npx -y agent-recall-mcpAdd AgentRecall to Cursor or VS Code
For Cursor, add the server to `.cursor/mcp.json`. For VS Code, add it to `.vscode/mcp.json`. The same npx command works for all clients.
# .cursor/mcp.json or .vscode/mcp.json
{
"mcpServers": {
"agent-recall": {
"command": "npx",
"args": ["-y", "agent-recall-mcp"]
}
}
}Start a session with /arstart to load context
At the beginning of each agent session, call `session_start` (or use `/arstart`) to load corrections, cross-project insights, and warnings from previous sessions into the agent's context.
Use remember and recall during the session
During work, use the `remember` tool to write facts, decisions, and learnings. The tool auto-routes entries to the appropriate memory layer (palace/rooms for facts, skills for procedures, corrections for rules). Use `recall` to perform hybrid BM25 + vector search over stored memory.
End the session with /arsave to compound memory
At the end of each session, call `session_end` (or `/arsave`) to save the journal entry, extract cross-project insights, and update the memory trajectory. Skipping this step means no learning compounds.
Use the CLI for scripted capture and recall
The agent-recall-cli can be used in scripts or hooks to capture Q&A pairs and recall on topics directly from the terminal.
npx agent-recall-cli capture "How do we handle rate limiting?" "Use exponential backoff with jitter, max 3 retries"
npx agent-recall-cli recall "rate limiting"AgentRecall Examples
Client configuration
Add AgentRecall to claude_desktop_config.json. Memory is stored in ~/.agent-recall/ and requires no environment variables.
{
"mcpServers": {
"agent-recall": {
"command": "npx",
"args": ["-y", "agent-recall-mcp"]
}
}
}Prompts to try
Use these prompts to exercise AgentRecall's memory tools during an agent session.
- "Load my session context and show any corrections or warnings"
- "Remember that we use Postgres not MySQL for this project"
- "Recall everything you know about our authentication setup"
- "Record a correction: always run tests before pushing to main"
- "Save this session and note that the API refactor is complete"Troubleshooting AgentRecall
The agent has no memory of previous sessions after restarting
Memory only persists if `session_end` (or `/arsave`) was called at the end of the previous session. Without this step, the session journal is not written and nothing compounds. Add `/arsave` as a habit at the end of every agent session.
recall returns empty results even though memory was saved
The hybrid search (BM25 + vector + RRF + Hopfield re-rank) reads from `~/.agent-recall/projects/<slug>/`. Verify that the correct project slug is active for the current session. If projects are misconfigured, memory may be written to a different slug than where recall is searching.
npx -y agent-recall-mcp fails with a network or registry error
Ensure you have internet access when first running the command, as npx needs to download the package from the npm registry. Once cached, it will run offline. If behind a corporate proxy, set the HTTPS_PROXY environment variable before running npx.
Frequently Asked Questions about AgentRecall
What is AgentRecall?
AgentRecall is a Model Context Protocol (MCP) server that ai session memory with think-execute-reflect quality loops — give your agent a brain that survives every session. built on the intelligent distance principle. It connects AI assistants to external tools and data sources through a standardized interface.
How do I install AgentRecall?
Follow the installation instructions on the AgentRecall GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with AgentRecall?
AgentRecall works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is AgentRecall free to use?
Yes, AgentRecall is open source and available under the MIT license. You can use it freely in both personal and commercial projects.
AgentRecall Alternatives — Similar Knowledge & Memory Servers
Looking for alternatives to AgentRecall? Here are other popular knowledge & memory servers you can use with Claude, Cursor, and VS Code.
MemPalace
★ 52.6kA local AI memory system that stores all conversations verbatim and organizes them into navigable structures. It provides 19 MCP tools for AI assistants to search and retrieve past decisions, debugging sessions, and architecture debates automatically
Kratos
★ 25.7k🏛️ Memory System for AI Coding Tools - Never explain your codebase again. MCP server with perfect project isolation, 95.8% context accuracy, and the Four Pillars Framework.
Context Mode
★ 15.4kAn MCP server that preserves LLM context by intercepting large data outputs and returning only concise summaries or relevant sections. It enables efficient sandboxed code execution, file processing, and documentation indexing across multiple programm
Memu
★ 13.7kMemory for 24/7 proactive agents like OpenClaw.
MemOS
★ 9.3kMemOS (Memory Operating System) is a memory management operating system designed for AI applications. Its goal is: to enable your AI system to have long-term memory like a human, not only remembering what users have said but also actively invoking, u
Everos
★ 5.4kBuild, evaluate, and integrate long-term memory for self-evolving agents.
Browse More Knowledge & Memory MCP Servers
Explore all knowledge & memory servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up AgentRecall in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use AgentRecall?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.