AXME
Your AI coding agent should remember your project. Here's how.
Developers re-explain project context every session. 10–20 minutes of setup per day, every day. Decisions made last week are invisible today. The agent repeats mistakes.
AXME Code gives Claude Code persistent memory across sessions: file maps, decisions, active tasks, and context loaded automatically at session start — no re-explaining required.
Developers re-explain project context every session. 10–20 minutes of setup per day. Decisions made last week are invisible today.
The morning context tax
Developers using Cursor, Claude Code, or other coding agents lose 10–20 minutes per session re-explaining architecture, open PRs, and team conventions. Decisions from last Tuesday live only in chat scrollback.
AXME Code persists oracle, decisions, memories, safety rules, backlog, and handoff under .axme-code/ — loaded with axme_context at session start via MCP.
Example: auth refactor across three days
Monday: team records Decision D-012 — all auth changes require integration tests in CI. Tuesday: new session loads context, sees handoff "finish middleware PR #88", continues without re-uploading the RFC. Wednesday: pattern memory reminds agent to isolate TestClient per test class after flake incident.
Teammate picks up Thursday with the same KB — not a verbal handoff.
Morning setup time
Without AXME Code
# 15 min: re-explain repo, paste files, recap decisions
With AXME Code
# axme_context(path="/repo") # oracle + handoff + decisions loaded
SOLUTION
How teams solve this with AXME.
Team rollout
Run axme-code setup with essential-safety presets. Add MCP config in Cursor (.cursor/mcp.json) or Claude Code marketplace. Require axme_context in project rules at session start; use begin_close / finalize_close before end of day.
See /integrations/cursor/ and /integrations/claude-code/ for host-specific setup.
Common questions
- Works for whole team?
- Shared .axme-code/ in repo (usually gitignored) or synced storage — atomic writes safe under concurrent use.
- vs CLAUDE.md only?
- Static files lack decisions with enforcement, handoff ritual, searchable memories, and safety YAML.
- Cursor and Claude Code together?
- Same knowledge base; hook-based blocking is strongest on Claude Code today.
Related reading
Deeper dives from the AXME blog.
How I Fixed Claude Code's Amnesia (Actually)
Claude Code forgets your project every session. CLAUDE.md helps, but it doesn't scale. I built an MCP server plugin that gives Claude Code persistent memory, decisions, and safety guardrails.
Read post →I Went on Vacation. My Agent Came Back to an Empty Project.
Five days off. Session handoff gone. Half a week of decisions forgotten. Here's the 'what was I doing yesterday' problem and why session close is a ritual that matters.
Read post →
Related capabilities
Related links
Ship your first durable agent — in under 10 minutes.
Free tier. No credit card. Self-host or hosted — your choice.