mcpField
M
memfleet

memfleet

Remote
memfleet.comMemory & Knowledge0Updated 12w ago
RepositoryVisit website

About memfleet

memfleet is a hosted platform for governed shared memory across AI agent fleets, for teams operating multiple agents. Contributions are judged before becoming shared memory, with scopes, provenance, and operator visibility built into the system.

Key features

  1. 01Judged contributions to shared memory
  2. 02Memory organized by scopes
  3. 03Provenance for every record entry
  4. 04Composed agent perspectives
  5. 05Operator controls in a web Console

Use cases

  1. 01Share reviewed lessons across agent sessions
  2. 02Provide agents with scope-specific memory
  3. 03Inspect what an agent can see
  4. 04Manage memory contributions and records

MCP support

The Streamable HTTP MCP server lets registered agents read perspectives and scopes, contribute or publish memory, and manage sessions. Agents authenticate with a bearer token issued through the Workspace Console; equivalent operations are also available through REST.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://memfleet.com/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

All endpoints
Streamable HTTPhttps://memfleet.com/mcpAuth required

Tools

Details

From the project README

memfleet

Shared, governed memory for fleets of AI agents. Contributions pass judgment; every entry carries provenance; operators see everything.

This repository is the server home for the memfleet MCP server — com.memfleet/memfleet on the official MCP Registry. The platform lives at memfleet.com; the command-line client ships as memfleet on PyPI.

What memfleet is

memfleet is the hosted platform for Strata, the open-source memory engine. You create a Workspace (a full memory fleet), register agents that receive keys, and operate everything from a web Console while your agents read and contribute over MCP or REST. Judgment runs as a platform service.

Agent teams forget: every session starts from zero and lessons learned by one agent never reach the rest. With memfleet, agents send contributions; nothing writes to shared memory directly. A judgment step decides what enters and whether it lands as a binding directive or informative context. Memory is organized in scopes arranged in strata, each agent reads a composed perspective of what its scope can see, and the record keeps provenance for every entry. Operators manage the whole lifecycle from a web Console, including seeing any agent's exact perspective.

The MCP server

  • Endpoint: https://memfleet.com/mcp (streamable HTTP)
  • Auth: Authorization: Bearer <agent key> — minted for a Registered Agent in your Workspace's Console; resolve it from an environment variable, never commit it.
  • Tools: read perspective, list scopes, read scope summary and record, contribute, publish, withdraw, session stats, session closeout.
  • REST: the same operations at /agent/v1 for anything that speaks HTTP.

Agents read and contribute over MCP or REST, so any MCP client can join the fleet; first-class setup ships for Claude Code today.

The server.json in this repository is the registry listing manifest for com.memfleet/memfleet.

Quickstart

pipx install memfleet
memfleet connect                                        # once per machine — one browser approval
memfleet run --profile 'Fleet/backend-bot' claude       # per project / per session

connect links the machine to your account. run binds the session to a Registered Agent — creating it if it does not exist yet — writes the project wiring, and launches the command. The setup it writes is secret-free: no key is ever written to disk in plaintext.

What the fleet gets

  • Governed memory, not a shared scratchpad. Every contribution passes judgment before it enters shared memory; every entry lives in a scope and carries provenance in the record.
  • Memory that doesn't eat your context window. An agent reads its perspective: a compact, composed summary of what its scope can see, not a transcript dump. That keeps the tokens your agent spends on memory small, and reading a perspective is never metered.
  • The operator is sovereign. See exactly what any scope's agents see, and steer memory directly. No hidden state.
  • The engine is open source. Run Strata yourself, or let memfleet run the fleet for you.

Links

License

MIT

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.