AgentRecall-MCP

Overview

AgentRecall is an MCP server that provides persistent memory for AI agents through a sophisticated system of journaling, awareness compounding, and semantic recall. It features Think-Execute-Reflect quality loops that allow agents to learn from previous interactions across sessions. The system indexes memories, applies weight and decay algorithms, and maintains a feedback loop for continuous improvement. It works with existing Git repos, Claude's memory, and other systems via a bootstrap import process.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• arstatusarstatus()

  Shows status board across all projects, showing pending work, blockers, relevance scores, and recommended next actions

  Try:Show me my current projects and what I should work on nextTry:What projects are blocked and why?Try:Display my project status board
• arsavearsave()

  Saves the current session by writing journal entries, consolidating insights, and updating palace memory

  Try:Save my work from this sessionTry:End the session and save all my insightsTry:Compound my learning from today's work
• arstartarstart(project: string)

  Loads deep context for a specific project including palace rooms, corrections, and task-specific recall

  Try:Start working on my project 'novada-mcp'Try:Load context for my 'agentrecall' projectTry:Begin session for project 'aam'
• arsaveallarsaveall()

  Batch saves all parallel sessions across projects, scanning, merging, deduplicating insights, and updating awareness in one operation

  Try:Save all my parallel sessions from todayTry:Batch save all open projectsTry:Consolidate and save all my work across sessions
• arbootstraparbootstrap(options?: { dryRun?: boolean, project?: string })

  Scans the machine for existing projects and imports them automatically, supporting migration from other memory systems

  Try:Import my existing projects from other toolsTry:Scan my machine for projects to importTry:Bootstrap from my Claude AutoMemory
• bootstrap_scanbootstrap_scan()

  Discovers existing projects and context on the machine without importing them

  Try:What projects would be imported if I bootstrap?Try:Scan for existing projects without importingTry:Show me what's ready to be imported
• bootstrap_importbootstrap_import(scan_result: any)

  Imports selected projects from a previous scan result

  Try:Import these specific projects from my scanTry:Bring in my 'agentrecall' and 'novada-mcp' projectsTry:Import just these three projects
• arcorrectionarcorrection(correction: string)

  Captures and stores corrections to improve agent behavior in future sessions

  Try:Remember to use Playwright instead of bb-browserTry:Note that I prefer light themes over dark onesTry:Capture this as a correction for future sessions
• arinsightarinsight(insight: string)

  Adds insights to the knowledge base that can be recalled across sessions and projects

  Try:Add this insight about server-rendered cards to my knowledgeTry:Remember that stemming improves keyword recallTry:Capture this as a reusable insight
• arprefetcharprefetch(query: string)

  Pre-fetches relevant memory before starting a new session or task

  Try:Find relevant memories about deploying to VercelTry:Prefetch context about multi-agent workflowsTry:Retrieve insights about async API patterns

Comparable tools

Installation

npm install -g agent-recall-mcp

{
  "mcpServers": {
    "agent-recall": {
      "command": "npx",
      "args": ["agent-recall-mcp"]
    }
  }
}

FAQ

What are the core commands in AgentRecall?

The two essential commands are `/arstatus` at the start of each session to select projects, and `/arsave` at the end to save session memory and compound insights. Additional commands include `/arstart` to load context, `/arsaveall` for batch saving, and `/arbootstrap` for importing existing memory systems.

Does AgentRecall work with other memory systems?

Yes, AgentRecall can automatically discover and import memory from Claude AutoMemory, Git repos, CLAUDE.md files, and other systems using the `/arbootstrap` command. It scans your machine and imports existing context without overwriting source files.

Compare AgentRecall-MCP with