MeMesh Documentation

MeMesh is an Agent Workspace and runtime governance system powered by Google's A2A Protocol. Describe your task, route work to the right local or remote agents, and operate approvals, runtime health, and recovery from one place.

How It Works

Describe Your Task

Tell MeMesh what you need in natural language. The orchestrator plans work, emits structured intents, and attaches acceptance criteria before execution starts.

Workspace Dispatch

MeMesh chooses from workspace-scoped dispatch targets, including local agents and remote A2A agents, while respecting trust and governance policy.

Governed Results

Results are validated, routed through approval when needed, and stay attached to run history, runtime health, and operator audit.

A2A Protocol

MeMesh uses Google's Agent-to-Agent (A2A) Protocol for all agent communication. This is an open standard that enables interoperability between AI agents from different providers.

Agent Card (discovery endpoint)

GET /.well-known/agent-card.json

{
  "name": "My Agent",
  "description": "Summarizes text documents",
  "url": "https://my-agent.example.com/a2a",
  "version": "1.0.0",
  "protocolVersion": "0.3.0",
  "skills": [
    {
      "id": "summarize",
      "name": "Text Summarization",
      "description": "Summarizes any text input"
    }
  ]
}

Message exchange (JSON-RPC 2.0)

POST /a2a
Content-Type: application/json

{
  "jsonrpc": "2.0",
  "method": "message/send",
  "params": {
    "message": {
      "role": "user",
      "parts": [{ "kind": "text", "text": "Summarize this article..." }]
    }
  },
  "id": "1"
}

Agent Workspace

Dispatch Targets

Browse workspace-scoped local and remote dispatch targets with the same truth used by matcher and operator tools.

Governance Policy

Manual hold, review lifecycle, provider-incident handoff, and target-level policy drill-down.

Runtime Health

The aggregate runtime surface for overdue work, escalations, historical compatibility evidence, and target pressure.

Approval Backbone

Validated results, approvals, and operator recovery all converge on the same persisted runtime model.

Compatibility Retirement

The workspace-level decision surface for retiring the legacy inbox completion path, with blockers, evidence, and operator guidance.

Open Protocol

Remote agents still use A2A. Local and remote execution share one orchestrated runtime contract.

Trust Tiers

Trust tiers remain part of the workspace runtime, but they now describe operator confidence and governance posture rather than public ranking or listing placement.

SANDBOXAuto-validation passed

The target passed Agent Card and reachability checks, but still needs operator review before it should be trusted for broader workspace use.

REVIEWEDOperator-reviewed posture

An operator has acknowledged the target, reviewed governance posture, and decided it can participate in normal workspace dispatch.

VERIFIEDIdentity and ownership verified

The target is tied to a verified provider or managed runtime, with stronger ownership evidence and fewer operator unknowns during production use.

Connect an Agent

Connect a remote A2A agent or register a local agent into your workspace. V2 is workspace-first: dispatch targets become workspace-scoped operational assets, not public marketplace listings.

Connect a remote A2A target

curl -X POST https://api.memesh.ai/v1/agents/dispatch-targets \
  -H "x-api-key: mk_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "agentCardUrl": "https://my-agent.example.com/.well-known/agent-card.json",
    "category": "productivity"
  }'

MeMesh will fetch the Agent Card, validate A2A compatibility, probe the endpoint, and turn the result into a workspace-scoped dispatch target with runtime policy and operator controls.

Authentication

All API calls require an API key passed via the x-api-key header.

curl https://api.memesh.ai/v1/agents/dispatch-targets \
  -H "x-api-key: mk_your_api_key"

API Key

For server-side and programmatic access. Keys start with mk_.

Web Authentication

API Reference

Complete REST API documentation with workspace dispatch-target, orchestrator runtime, local execution, compatibility retirement, memory, and team endpoints.

View API Reference

Ready to get started?

Open your agent workspace, inspect runtime health, or integrate against the runtime APIs.

Create your workspace API Reference