Deployment¶
Deployment Options¶
github-agent exposes its MCP server (console script github-mcp) four ways. Pick the row that
matches where the server runs relative to your MCP client, then copy the matching
mcp_config.json below. Replace the <your-…> placeholders with the values from the Configuration / Environment Variables section.
| # | Option | Transport | Where it runs | mcp_config.json key |
|---|---|---|---|---|
| 1 | stdio | stdio |
client launches a subprocess | command |
| 2 | Streamable-HTTP (local) | streamable-http |
a local network port | command or url |
| 3 | Local container / uv | stdio or streamable-http |
Docker / Podman / uv on this host | command or url |
| 4 | Remote URL | streamable-http |
a remote host behind Caddy | url |
1. stdio (local subprocess)¶
The client launches the server over stdio via uvx — best for local IDEs
(Cursor, Claude Desktop, VS Code):
{
"mcpServers": {
"github-mcp": {
"command": "uvx",
"args": ["--from", "github-agent", "github-mcp"],
"env": {
"GITHUB_URL": "<your-github_url>"
}
}
}
}
2. Streamable-HTTP (local process)¶
Run the server as a long-lived HTTP process:
uvx --from github-agent github-mcp --transport streamable-http --host 0.0.0.0 --port 8000
curl -s http://localhost:8000/health # {"status":"OK"}
Then either let the client launch it:
{
"mcpServers": {
"github-mcp": {
"command": "uvx",
"args": ["--from", "github-agent", "github-mcp", "--transport", "streamable-http", "--port", "8000"],
"env": {
"TRANSPORT": "streamable-http",
"HOST": "0.0.0.0",
"PORT": "8000",
"GITHUB_URL": "<your-github_url>"
}
}
}
}
…or connect to the already-running process by URL:
3. Local container / uv¶
(a) Launch a container directly from mcp_config.json (stdio over the container —
no ports to manage). Swap docker for podman for a daemonless runtime:
{
"mcpServers": {
"github-mcp": {
"command": "docker",
"args": [
"run", "-i", "--rm",
"-e", "TRANSPORT=stdio",
"-e", "GITHUB_URL=<your-github_url>",
"knucklessg1/github-agent:latest"
]
}
}
}
(b) Run a local streamable-http container, then connect by URL:
docker run -d --name github-mcp -p 8000:8000 \
-e TRANSPORT=streamable-http \
-e PORT=8000 \
-e GITHUB_URL="<your-github_url>" \
knucklessg1/github-agent:latest
# or, from a clone of this repo:
docker compose -f docker/mcp.compose.yml up -d
(c) From a local checkout with uv:
4. Remote URL (deployed behind Caddy)¶
When the server is deployed remotely (e.g. as a Docker service) and published through
Caddy on the internal *.arpa zone, connect with the "url" key — no local process or
image required:
Caddy reverse-proxies http://github-mcp.arpa to the container's :8000
streamable-http listener; http://github-mcp.arpa/health returns
{"status":"OK"} when the service is live.
This page covers running github-agent as a long-lived service: the transports, a
Docker Compose stack, the optional A2A agent server, putting it behind a Caddy
reverse proxy, and giving it a DNS name with Technitium.
github-agentships two console scripts: an MCP server (github-mcp) — a typed, deterministic tool surface a policy router / agent calls — and an A2A agent server (github-agent) that delegates GitHub work to specialized child agents. Deploy the MCP server alone, or both together.
Run the MCP server¶
The transport is selected with --transport (or the TRANSPORT env var):
Health check (HTTP transports):
Configuration (environment)¶
github-agent is configured entirely from the environment. The required set for
the MCP server:
| Var | Default | Meaning |
|---|---|---|
GITHUB_URL |
https://api.github.com |
GitHub REST API base URL (set for GitHub Enterprise) |
GITHUB_TOKEN |
unset | Personal access token / GitHub App token (Bearer) |
GITHUB_VERIFY |
True |
Verify TLS (set False only for self-signed Enterprise) |
HOST |
0.0.0.0 |
Bind address for HTTP transports |
PORT |
8000 |
Bind port for HTTP transports |
TRANSPORT |
stdio |
stdio, streamable-http, or sse |
Each tool domain is gated by its own switch — REPOTOOL, ISSUETOOL, PULLTOOL,
CONTENTTOOL, BRANCHTOOL, COMMITTOOL, SEARCHTOOL, ORGTOOL,
COLLABORATORTOOL, ACTIONTOOL, RELEASETOOL (all default True). The full set,
including telemetry (ENABLE_OTEL, OTEL_EXPORTER_OTLP_*) and access governance
(EUNOMIA_TYPE), is documented in
.env.example.
Copy it to .env and populate only what you use; an absent GITHUB_TOKEN leaves the
client unauthenticated rather than raising.
Backing service¶
GitHub is a managed (SaaS) service — there is no backing system to deploy. Point
GITHUB_URL at https://api.github.com for GitHub.com, or at your GitHub Enterprise
Server REST endpoint (for example, https://github.example.com/api/v3), and supply a
GITHUB_TOKEN. Only connection configuration is required.
Docker Compose¶
The repo ships docker/mcp.compose.yml.
It reads a sibling .env and publishes the HTTP server on :8000:
services:
github-agent-mcp:
image: knucklessg1/github-agent:latest
container_name: github-agent-mcp
hostname: github-agent-mcp
restart: always
env_file:
- ../.env
environment:
- PYTHONUNBUFFERED=1
- HOST=0.0.0.0
- PORT=8000
- TRANSPORT=streamable-http
ports:
- "8000:8000"
healthcheck:
test: ["CMD", "python3", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8000/health')"]
interval: 30s
timeout: 10s
retries: 3
cp .env.example .env # then set GITHUB_TOKEN and GITHUB_URL
docker compose -f docker/mcp.compose.yml up -d
docker compose -f docker/mcp.compose.yml logs -f
Agent server (A2A)¶
The A2A supervisor agent is published as the github-agent console script and
docker/agent.compose.yml,
which provisions the MCP server and the agent together. The agent reaches the MCP
server over MCP_URL and publishes its own HTTP API (and optional web UI) on :9016:
services:
github-agent-mcp:
image: knucklessg1/github-agent:latest
hostname: github-agent-mcp
env_file: [../.env]
environment:
- TRANSPORT=streamable-http
- HOST=0.0.0.0
- PORT=8000
ports: ["8000:8000"]
github-agent-agent:
image: knucklessg1/github-agent:latest
depends_on: [github-agent-mcp]
command: ["github-agent"]
env_file: [../.env]
environment:
- HOST=0.0.0.0
- PORT=9016
- MCP_URL=http://github-agent-mcp:8000/mcp
- PROVIDER=${PROVIDER:-openai}
- MODEL_ID=${MODEL_ID:-gpt-4o}
- ENABLE_WEB_UI=True
ports: ["9016:9016"]
docker compose -f docker/agent.compose.yml up -d
curl -s http://localhost:9016/health # agent health
Configure the agent's model with PROVIDER, MODEL_ID, and the matching provider
API key (for example LLM_API_KEY / OPENAI_API_KEY).
Behind a Caddy reverse proxy¶
Expose the HTTP server on a hostname with automatic TLS. Add to your Caddyfile:
# Internal (self-signed) — homelab .arpa zone
github-agent.arpa {
tls internal
reverse_proxy github-agent-mcp:8000
}
Reload Caddy:
DNS with Technitium¶
Point the hostname at the host running Caddy. Via the Technitium API:
curl -s "http://technitium.arpa:5380/api/zones/records/add" \
--data-urlencode "token=$TECHNITIUM_DNS_TOKEN" \
--data-urlencode "domain=github-agent.arpa" \
--data-urlencode "zone=arpa" \
--data-urlencode "type=A" \
--data-urlencode "ipAddress=10.0.0.10" \
--data-urlencode "ttl=3600"
…or add an A record github-agent.arpa → <caddy-host-ip> in the Technitium web
console (http://technitium.arpa:5380). The ecosystem
technitium-dns-mcp automates
this as a tool.
Register with an MCP client¶
Add to your client's mcp_config.json (multiplexer nickname gh):
{
"mcpServers": {
"github-agent": {
"command": "uv",
"args": ["run", "github-mcp"],
"env": {
"MCP_TOOL_MODE": "condensed",
"GITHUB_URL": "https://api.github.com",
"GITHUB_TOKEN": "ghp_your_personal_access_token",
"REPOTOOL": "True",
"ISSUETOOL": "True",
"PULLTOOL": "True"
}
}
}
}
For a remote HTTP server, point the client at http://github-agent.arpa/mcp instead.