Skip to main content
Add AMFS to Cursor, Claude Code, or any MCP client. Your agents get persistent memory that compounds across sessions, machines, and teammates.

​
Cursor

Add to .cursor/mcp.json (project-level) or ~/.cursor/mcp.json (global): 
{
  "mcpServers": {
    "amfs": {
      "command": "uvx",
      "args": ["amfs-mcp-server"],
      "env": {
        "AMFS_HTTP_URL": "https://amfs-login.sense-lab.ai",
        "AMFS_API_KEY": "<your-api-key>"
      }
    }
  }
}
Copy the ready-made snippet from your AMFS Dashboard → Agents page. The MCP Connection Card has your URL and key pre-filled.

​
Claude Code

Add to ~/.claude/claude_desktop_config.json: 
{
  "mcpServers": {
    "amfs": {
      "command": "uvx",
      "args": ["amfs-mcp-server"],
      "env": {
        "AMFS_HTTP_URL": "https://amfs-login.sense-lab.ai",
        "AMFS_API_KEY": "<your-api-key>"
      }
    }
  }
}

​
Any MCP Client

Same pattern — set AMFS_HTTP_URL and AMFS_API_KEY as env vars on the uvx amfs-mcp-server process. No local code, no cloning, no Docker.

​
Verify

Restart your IDE after saving the config. Ask the agent to run amfs_stats() — if it returns a response, you’re connected.

​
What Your Agents Get

Once connected, agents have 20+ tools across five categories: Identity & Context
ToolWhat it does
amfs_set_identityName this agent session (e.g. "checkout-fixer")
amfs_briefingGet a compiled knowledge digest — what matters right now
Read & Write
ToolWhat it does
amfs_read / amfs_writeRead and write versioned memory entries
amfs_searchFilter by entity, confidence, agent, time range
amfs_retrieveNatural language search with semantic + recency + confidence ranking
amfs_list / amfs_statsBrowse entries and get an overview
amfs_historyVersion history of any entry
amfs_graph_neighborsExplore the knowledge graph
Agent Brain
ToolWhat it does
amfs_recallRecall your own memories
amfs_my_entriesList everything you’ve written
amfs_read_fromRead from another agent’s memory (tracked)
Decision Traces
ToolWhat it does
amfs_record_contextCapture tool results, user choices, external data
amfs_commit_outcomeSnapshot the full decision trace
amfs_explainInspect the causal chain
Timeline
ToolWhat it does
amfs_timelineBrowse the git-style event log

​
Agent Rules

To get the most out of AMFS, add the agent rules file to your project. This teaches agents when and how to use memory tools automatically: Cursor — copy amfs-memory.mdc to .cursor/rules/amfs-memory.mdc in your project. Claude Code — copy CLAUDE.md to your project root. Without rules, agents still have access to the tools — they just won’t use them proactively.

​
Agent Identity

Agents should call amfs_set_identity at the start of every session: 
amfs_set_identity("checkout-agent", "Fixing retry logic")
Use kebab-case role names that persist across conversations:
GoodBad
dashboard-agentfix-button-color (too specific)
api-agentagent-1 (meaningless)
If amfs_set_identity isn’t called, the MCP server auto-detects identity from the IDE environment (cursor/<username>, claude-code/<username>).

​
How Knowledge Compounds

Machine A (Cursor / Bruno):
  → Reviews checkout-service PR
  → amfs_write("myapp/checkout", "risk-race-condition", "...")
  → amfs_commit_outcome("PR-456", "minor_failure")

Machine B (Claude Code / Alice):
  → Starts working on checkout-service
  → amfs_briefing(entity_path="myapp/checkout")
  → Sees Bruno's risk signal with boosted confidence
  → Avoids the same issue

​
Pinning a Version

To pin to a specific MCP server version: 
"args": ["amfs-mcp-server@0.2.3"]

​
Troubleshooting

ProblemFix
uvx: command not foundInstall uv: curl -LsSf https://astral.sh/uv/install.sh | sh
Agent doesn’t use memory toolsAdd the agent rules file to your project
401 UnauthorizedCheck your API key in Dashboard → Settings → API Keys
403 Scope DeniedYour key doesn’t have permission for that entity path