AgentMesh
AgentMesh is an open-source registry and discovery service for AI agents.
It provides a standard AgentCard model, REST API, SDK, CLI, optional token auth, and pluggable storage backends.
What is implemented
- Agent registration, update, deletion, listing
- Agent discovery (skill/protocol/tags/query) and search
- Protocol invocation gateway (
http,custom,a2a,mcp,grpc,websocket) - Heartbeat + health checks
- System and per-agent statistics
- Token endpoints (
/auth/token,/auth/refresh,/auth/verify) - Optional API key protection for write/admin routes
- Production safety mode (
--production) to require strong auth config - Signature utilities (
/security/keypair,/security/sign,/security/verify) - Storage backends: memory (default), Redis, PostgreSQL
- Python SDK: async
AgentMeshClient+ syncSyncAgentMeshClient - CLI:
serve,config,agentscommands
Install
pip install -e .
Optional protocol extras (for grpc/websocket invocation bridges):
pip install -e ".[grpc,websocket]"
Quick start
1) Run server
# in-memory (default)
python -m agentmesh serve --storage memory --port 8000
# Redis
python -m agentmesh serve --storage redis --redis-url redis://localhost:6379 --port 8000
# PostgreSQL
python -m agentmesh serve --storage postgres --postgres-url postgresql://localhost:5432/agentmesh --port 8000
# production safety checks enabled
python -m agentmesh serve --storage postgres --postgres-url postgresql://localhost:5432/agentmesh \
--api-key YOUR_API_KEY --auth-secret YOUR_STRONG_SECRET --production --port 8000
2) Register and discover with SDK
import asyncio
from agentmesh import AgentMeshClient
async def main():
client = AgentMeshClient(base_url="http://localhost:8000")
await client.register_agent({
"id": "weather-bot-001",
"name": "WeatherBot",
"version": "1.0.0",
"description": "Weather forecasting service",
"skills": [
{"name": "get_weather", "description": "Get current weather"}
],
"endpoint": "http://localhost:8001/weather",
"protocol": "http",
"tags": ["weather", "api"],
"health_status": "healthy"
})
result = await client.discover_agents(skill="get_weather")
print(result["data"]["agents"])
invoke_result = await client.invoke_agent(
"weather-bot-001",
payload={"city": "Tokyo"},
path="/weather"
)
print(invoke_result["data"]["result"]["response"])
await client.close()
asyncio.run(main())
3) Web Dashboard (EvoMap UI)
AgentMesh now includes a high-fidelity web console inspired by evomap.ai.
# Register some demo agents first
python seed_registry.py
# Start backend
python -m agentmesh serve --debug
# Start frontend (in a new terminal)
cd web
npm install
npm run dev
Visit http://localhost:3000 to view analytics, search agents (⌘K), and test Manifests in the Sandbox.
4) Use CLI
# configure endpoint
agentmesh config set endpoint http://localhost:8000
# register
agentmesh agents register \
--id cli-agent-001 \
--name CLIAgent \
--description "CLI agent" \
--skill execute_command \
--tag cli
# list/search/get/update/delete
agentmesh agents list
agentmesh agents search --skill execute_command
agentmesh agents get cli-agent-001
agentmesh agents invoke cli-agent-001 --payload '{"task":"ping"}'
agentmesh agents update cli-agent-001 --description "Updated description"
agentmesh agents delete cli-agent-001
🔒 Production Readiness
The server includes a --production mode that enforces security best practices:
agentmesh serve --production \
--api-key your-production-key \
--auth-secret your-strong-jwt-secret \
--storage postgres \
--postgres-url postgresql://user:pass@db:5432/agentmesh
What production mode enforces:
X-API-Keyis mandatory for all write operations.auth-secretmust be provided and cannot be the default value.- Rate limiting is enabled on sensitive endpoints.
- Prometheus
/metricsare exposed for observability.
API docs
- Swagger:
http://localhost:8000/docs - ReDoc:
http://localhost:8000/redoc
Detailed reference: docs/protocol/api_reference.md.
Response format
Successful responses:
{
"success": true,
"data": {},
"message": "Operation successful",
"timestamp": "2026-02-23T18:00:00Z"
}
Error responses:
{
"success": false,
"error": {
"code": "400",
"message": "Validation failed",
"details": {}
},
"timestamp": "2026-02-23T18:00:00Z"
}
Run tests
python -m unittest discover -s tests -v
Project layout
src/agentmesh/
├── __init__.py
├── __main__.py
├── cli.py
├── client.py
├── api/
│ ├── routes.py
│ └── server.py
├── auth/
│ └── token_manager.py
├── core/
│ ├── agent_card.py
│ ├── registry.py
│ └── security.py
├── storage/
│ ├── base.py
│ ├── memory.py
│ ├── redis.py
│ └── postgres.py
├── protocols/
│ ├── base.py
│ ├── gateway.py
│ ├── http_custom.py
│ ├── a2a.py
│ ├── mcp.py
│ ├── grpc.py
│ └── websocket.py
└── utils/
└── responses.py
License
Apache-2.0
Metadata
Release files for agentmesh-python 1.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| agentmesh_python-1.1.0.tar.gz | 126.9 kB | Without provenance | Details for agentmesh_python-1.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| agentmesh_python-1.1.0-py3-none-any.whl | Python 3 | none | any | Details for agentmesh_python-1.1.0-py3-none-any.whl |
Total release size: 230.0 kB
Release files / agentmesh_python-1.1.0.tar.gz
| Download URL | agentmesh_python-1.1.0.tar.gz |
|---|---|
| Size | 126.9 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
5d50050ae90f3e81ebf0d61a05ff87e44ab4edc56614fe9dc9adc554bde5e0db
|
|
BLAKE2b-256 checksum How to use checksums |
081f888eef0cee47a20e47da1326560e82751d502235202b005ff64b853de460
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.14.3
|
Release files / agentmesh_python-1.1.0-py3-none-any.whl
| Download URL | agentmesh_python-1.1.0-py3-none-any.whl |
|---|---|
| Size | 103.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
40f3c09686c77d0aa3669f8d8e1b2d4ead549cb9a2809f365c1d9359f43473c1
|
|
BLAKE2b-256 checksum How to use checksums |
f0acfead5d8ab7742ea9cc04edb00546735f46ef559cd283cbf82a77e774b3e7
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.14.3
|
Security for agentmesh-python 1.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.