# A2CR
A2CR is not an AI.
It is the baton that lets AI agents hand work to one another.
A2CR is an MCP-first work-continuation layer.
It lets MCP-capable tools save focused WorkBaton checkpoints and resume work later from another window, model, or AI agent configured with A2CR MCP.
Long AI sessions get heavy because the work history grows. Later turns often need earlier requirements, decisions, files, errors, and corrections, so the active context can become slower and more token-hungry.
A2CR keeps context light. When work reaches a milestone or context pressure appears, an agent can save a focused WorkBaton with the goal, current state, next action, blockers, and validation. Pro has a larger size budget, so it can carry a richer handoff while still moving bulky support notes to WorkStash. A fresh AI session can resume from that distilled state instead of carrying the entire chat history forward.
Important URLs:
- Guide: https://a2cr.app/en/guide
- Manual: https://a2cr.app/en/manual
- AI agent guide: https://a2cr.app/en/agent-guide
- Pricing: https://a2cr.app/pricing
- MCP service URL: https://a2cr.app/mcp
- LLM notes: https://a2cr.app/llms.txt
Public setup:
- Install or update the local stdio MCP wrapper with python -m pip install --upgrade a2cr-mcp.
- Register the installed a2cr-mcp command as one MCP server named a2cr.