epik-gh
A GitHub MCP server that wraps the gh CLI to expose GitHub operations like issues, pull requests, branches, labels, repositories, CI actions, and Projects V2 as tools for MCP clients.
README
EpikMCP
EpikMCP is the single MCP for Epik. It exposes its functionality through two internal modules:
- plan — GitHub access via the
ghCLI. Read and write issues, relationships, projects, labels and repos, plus read-only access to pull requests and CI runs. - build — launch feature builds via Anthropic. Fires a saved Claude Code routine to start a remote feature build.
An MCP client (such as Claude) calls these tools directly. Tools are registered
under the mcp__epik-mcp__* prefix.
Module and tool layout
plan / GitHub (via gh)
All plan tools run through the gh CLI, so they operate with whatever GitHub
account gh auth login has authenticated.
- Issues (
issues)issue_list— list issues in a repositoryissue_get— get a single issueissue_create— create an issueissue_edit— edit an issueissue_close— close an issueissue_reopen— reopen an issueissue_comment— comment on an issue
- Relationships (
relationships)issue_set_blocked_by— mark an issue as blocked by anotherissue_remove_blocked_by— remove a blocked-by relationshipissue_list_relationships— list an issue's relationshipsissue_add_sub_issue— add a sub-issueissue_remove_sub_issue— remove a sub-issue
- Projects V2 (
projects)project_list_items— list project itemsproject_get_item— get a single project itemproject_set_status— set an item's statusproject_invalidate_cache— invalidate the cached project IDs
- Labels (
labels)label_list— list labelslabel_create— create a labellabel_delete— delete a label
- Repositories (
repos)repo_get— get repository metadatarepo_default_branch— get the default branch
- Pull requests (
prs, read-only)pr_list— list pull requestspr_get— get a single pull request
- CI / Actions runs (
runs, read-only)run_list— list workflow runsrun_get— get a single runrun_logs— fetch run logs
- Raw passthrough (
raw)gh_raw— run a rawghsubcommand
- Feature status (
feature_status)feature_status— aggregate the plan-side status of a feature
build / Anthropic
- Feature launch (
feature_launch)feature_launch— start a remote feature build
See Build module: feature_launch below for the
one-time routine setup and required environment variables.
Prerequisites
- Python 3.11+
uv(recommended) or pipghCLI installed and authenticated (gh auth login) for the plan module
Installation
Run directly with uvx
uvx --from git+https://github.com/wpm/EpikMCP.git epik-mcp
Install as a uv tool
uv tool install git+https://github.com/wpm/EpikMCP.git
Or clone and install locally:
git clone https://github.com/wpm/EpikMCP.git
cd EpikMCP
uv tool install .
With pip
pip install git+https://github.com/wpm/EpikMCP.git
Configuring as a CoWork / Claude MCP server
Add the following to your Claude MCP config (for example,
~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"epik-mcp": {
"command": "uvx",
"args": [
"--from",
"git+https://github.com/wpm/EpikMCP.git",
"epik-mcp"
],
"env": {
"EPIK_ROUTINE_ID": "<your-routine-id>",
"EPIK_ROUTINE_TOKEN": "<your-routine-token>"
}
}
}
}
If you installed the epik-mcp command as a uv tool instead, you can point the
config directly at the binary:
{
"mcpServers": {
"epik-mcp": {
"command": "/Users/YOUR_USERNAME/.local/bin/epik-mcp",
"env": {
"EPIK_ROUTINE_ID": "<your-routine-id>",
"EPIK_ROUTINE_TOKEN": "<your-routine-token>"
}
}
}
}
Confirm the binary path with:
which epik-mcp
The build-module env vars are only needed if you intend to use feature_launch.
Authentication
EpikMCP keeps the two modules' authentication separate:
-
plan module delegates all authentication to the
ghCLI. Before using the plan tools, make sure you are logged in:gh auth loginTo verify:
gh auth status -
build module uses the routine environment variables (
EPIK_ROUTINE_IDandEPIK_ROUTINE_TOKEN) described below. It does not use theghCLI.
Build module: feature_launch
The build module is the Anthropic-side half of Epik. It keeps its auth and state
separate from the gh CLI used by the plan module. The feature_launch
tool starts a remote feature build by firing a saved
Claude Code routine via the routines API.
When called with a feature issue number plus a base and target branch, it POSTs to the routine's fire endpoint and returns the URL of the cloud session it started.
One-time routine setup
- In the Claude Code web UI, create a routine (for example, a "feature runner") that builds a feature when given a feature command.
- Add an API trigger to the routine. Its prompt should be the feature-command
body (the instruction your build follows).
feature_launchsends the feature issue number and branches as the trigger'stext. - Enable "Allow unrestricted branch pushes" for the repository so the routine can push the build branch.
- Copy the routine's id and its API token — you'll set these as env vars.
Required environment variables
These are read on every call and must be set in the environment the MCP server runs in:
| Variable | Description |
|---|---|
EPIK_ROUTINE_ID |
The Claude Code routine id whose fire endpoint is called. |
EPIK_ROUTINE_TOKEN |
The routine API bearer token. |
If either is missing or empty, feature_launch raises a clear validation error
naming the missing variable. Non-2xx responses and connection failures also raise
clear errors.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.