Recall
Give your coding agent durable, searchable memory across sessions instead of re-explaining your stack every time.
Overview
Recall is an MCP server for the Build phase that provides persistent, searchable, tiered agent memory across sessions.
What is this MCP server?
- Open-source MCP server (RecallWorks) with stdio transport via PyPI package ai-recallworks
- Persistent, searchable, tiered memory designed to survive across chat sessions
- Version 0.5.2 in the official MCP server schema
- Self-hosted memory layer—no proprietary cloud lock-in implied by the listing
- Fits solo builders who ship with agents daily and need context continuity
- Server version 0.5.2
- PyPI identifier ai-recallworks
- Transport: stdio
Community signal: 3 GitHub stars.
What problem does it solve?
Agents forget prior session context, so solo builders waste tokens re-stating architecture, prefs, and half-finished decisions.
Who is it for?
Solo builders running long Claude Code or Cursor sessions who want self-hosted MCP memory instead of ad-hoc notes.
Skip if: Teams that only need in-repo documentation with no cross-session agent recall, or builders avoiding extra MCP processes.
What do I get? / Deliverables
After you register Recall, your agent can query and retain structured memory so multi-day builds stay coherent without manual copy-paste.
- Registered stdio MCP memory server
- Searchable cross-session memory store for agent tools
- Tiered retention model per Recall server design
Recommended MCP Servers
Journey fit
Agent memory is installed and wired during product build when you configure MCP tooling for Claude Code, Cursor, or similar clients. Recall is explicitly an MCP memory server—canonical placement is agent-tooling, not generic backend or docs.
How it compares
MCP memory server integration, not a brainstorming or planning agent skill.
Common Questions / FAQ
Who is Recall for?
Recall is for indie and solo developers who use MCP-capable agents and need memory that persists and searches across sessions.
When should I use Recall?
Use Recall during Build when you configure agent-tooling MCP servers and want tiered, searchable context instead of one-shot chat history.
How do I add Recall to my agent?
Install the PyPI package ai-recallworks (stdio), add the server entry to your Claude Code, Cursor, or Codex MCP config, and restart the client.