Prototype · Preview: every server in the official registry, synced every 30 minutes. Reviews and accounts come in a later step.
PopNx MCPMCP servers for people who use AI · explained in plain words

Self-hosted GitHub MCP server

by adelaidasofiaversion 0.1.0i★ –iRuns on your computer
Registry nameio.github.adelaidasofia/github-mcp

Install

Overview

Description from the maker (original English)

Self-hosted GitHub MCP server. PAT auth, stdio transport, transport-agnostic so it works with…

PopNx MCP’s three pillars

These three are never merged into one score: opinions, measurements and popularity are different things.

What you can do with it

We have not written a plain-language summary for this one yet. Below are the facts we read from the registry.

  • Connection: Runs on your computer
  • Transport: stdio
  • Category: Developer tools
  • You need to fill in: GITHUB_TOKEN

Install

Pick the AI app you use, then copy the settings.

Before you install

  • Basic checks: 1 of 1 passed
  • Runs on your computer with your user’s permissions (files, network, apps).
  • You fill in yourself: GITHUB_TOKEN

Values marked secret are keys to your account. Never share them or paste them into a chat.

Which app do you use?

Pick your app above. We remember it, so next time the button says “Install in …”.

Download the maker’s .mcpb file

The link goes to the maker’s own file; we do not host or repackage it. Open the file and Claude Desktop installs it.

Run this in your terminal

Fill in the empty values before running it.

Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.

Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.

This one runs on your computer, and ChatGPT can only connect to servers on the internet. Pick another app.

Your choice stays on this device in a cookie; no account needed.

Values you fill in

  • GITHUB_TOKEN secret required env
    Classic or fine-grained PAT with repo + workflow scopes

These settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.

What it can do

Runs on your computer; the tool list appears in your AI app after you install it.

Basic checksi

We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 6, 2026

  • ✓ Pass

    Maker matches the repo

    The registry name was verified with the GitHub account adelaidasofia, which owns the repo

  • … Not checked yet

    Still maintained

    We have not read this repo from GitHub yet

  • … Not checked yet

    Has a licence

    We have not read this repo from GitHub yet

  • – Not applicable

    Endpoint responds

    This one runs on your computer; there is no internet endpoint

  • – Not applicable

    Sign-in with OAuth

    Runs on your computer and uses local settings instead

  • … Not checked yet

    No hidden instructions in tool descriptions

    Runs on your computer; we can only read tool lists from internet endpoints

Compared with others in Developer toolsi

  • More stars than 0% of the other 1388 servers in this category
  • Published to the registry more recently than 22% of the other 1388 servers in this category
  • Basic checks equal to or better than 100% of the other 1388 servers in this category

Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.

Common questions

Is Self-hosted GitHub MCP server safe?

It passed 1 of 1 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.

What keys or codes does it need?

You need to fill in: GITHUB_TOKEN

Does it run on my computer or in the cloud?

It runs on your computer. Your AI app starts it when needed, and data does not pass through the maker’s servers (unless it calls an online service itself).

Is it still maintained?

The GitHub repo last changed –, and the latest version (0.1.0) reached the registry 6 months ago.

Which apps can use it?

Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. You also need MCPB on your computer.

What can I use instead?

In the same category: GitHub, Desktop Commander, Cloudflare MCP

Alternatives

Servers in the same category from other makers, sorted by stars.

Reading a card:★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi

GitHub

by github

GitHub’s official server lets AI read and manage code, issues and pull requests in your account. Use GitHub’s hosted version or run it locally with Docker.

Computer or cloud★ 34,245MITUpdated 3 days ago
Install

Desktop Commander

by wonderwhy-er

MCP server for terminal commands, file operations, and process management

Runs on your computer★ 10,225MITUpdated 15 days ago
Install

Cloudflare MCP

by cloudflare

Cloudflare MCP servers

Runs in the cloud★ 4,446Apache-2.0Updated 14 months ago
Install

README from the maker

Open the README

Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.

github-mcp

Self-hosted GitHub MCP server. Personal-access-token auth, stdio transport, transport-agnostic so it works with Claude Code (any account), Codex CLI, Cursor, or any MCP-compliant client. No claude.ai OAuth dependency.

Written in Python on top of FastMCP. Slots into the same install pattern as other community Python MCP servers.

Why this exists

The official github/github-mcp-server (Go, 60+ tools) is excellent if you want a kitchen-sink server tied to a single agent platform's auth flow. This one is scoped for operators who:

  • Switch between multiple Claude accounts, Codex CLI, Cursor, or other MCP clients and want the same GitHub auth across all of them.
  • Prefer a small, audited tool surface scoped to PR + repo file + Dependabot operations.
  • Want a Python install that lives alongside their other MCP servers.

If neither applies, run github/github-mcp-server instead — that's the right call.

Tool surface (v0.1.0)

ToolPurpose
health_checkVerify the PAT is valid + report rate-limit headroom. First call when wiring up.
list_prsList pull requests with state filter.
get_prSingle PR with check-run status + mergeable state.
merge_prSquash/merge/rebase with optional branch delete.
list_dependabot_alertsAlert sweep with severity-grouped counts.
get_fileRead a file via the Contents API.
put_fileCreate or update a file via the Contents API (idempotent with SHA).

Future versions expand: workflows, releases, issues, rulesets, secret-scan alerts, branch protection, repo settings.

Secret hygiene (read this first)

v0.1.1+ reads the PAT from macOS Keychain by default. Env-var fallback is still supported for non-macOS / CI, but the Keychain path is the recommended setup because the secret never enters argv or process env — safe from ps aux, pgrep -fl, lsof, /proc/ /environ, panic dumps, CI logs, debuggers, or any LLM agent running an introspection command.

# One-time setup (macOS):

security add-generic-password -s github-mcp -a "$USER" -w " "

Update later (e.g. after rotation):

security delete-generic-password -s github-mcp -a "$USER" 2>/dev/null security add-generic-password -s github-mcp -a "$USER" -w " "

Do NOT install with an inline --env:

# BAD — bakes the PAT into ~/.claude.json + every Claude child spawn's --mcp-config argv.

Any pgrep / ps aux dumps it.

claude mcp add github --scope user --env GITHUB_TOKEN=github_pat_... --

GOOD — no --env block. The server reads from Keychain at call time.

claude mcp add github --scope user --

PR/incident lineage: this hygiene path was added 2026-05-19 after a pgrep -fl chrome-devtools-mcp dumped a PAT-bearing claude process into a session transcript. See adelaidasofia/github-mcp Keychain migration (TODO: link PR once merged).

For non-macOS environments, fall back to env:

export GITHUB_TOKEN=<your-PAT>   # set in shell rc, NOT in argv

Minimum scopes for the v0.1 tool surface: Contents R/W, Pull requests R/W, Dependabot R.

Install

Open Claude Code, paste:

/plugin marketplace add adelaidasofia/github-mcp /plugin install github-mcp@github-mcp

Legacy install

git clone https://github.com/adelaidasofia/github-mcp ~/.claude/github-mcp

cd ~/.claude/github-mcp uv tool install --editable .

Or as a one-off without persistent install:

cd ~/.claude/github-mcp

uv run github-mcp

Auth

The server reads the GitHub PAT from environment, in this precedence:

  1. GITHUB_TOKEN
  2. GH_TOKEN

Fine-grained tokens recommended.

Scope matrix
PermissionAccessUnlocks
MetadataReadBaseline (auto-required when any other repo permission is set)
ContentsRead and writeget_file, put_file
Pull requestsRead and writelist_prs, get_pr, merge_pr
Dependabot alertsReadlist_dependabot_alerts
AdministrationRead and writeFuture: rulesets, branch protection, allow_auto_merge toggle
WorkflowsRead and writeFuture: read/update .github/workflows/*.yml
ActionsRead and writeFuture: trigger and inspect workflow runs
Secret scanning alertsReadFuture: fleet-wide secret-scan sweep
Code scanning alertsReadFuture: CodeQL fleet sweep

v0.1 floor: Metadata + Contents (R/W) + Pull requests (R/W) + Dependabot alerts (R) is enough to use every tool in this release.

Future-proof: granting the full table now means new MCP tools can ship without rotating the PAT. Equivalent to checking "all repository permissions" in the fine-grained PAT UI. Trade-off: broader blast radius if the token leaks. Reasonable for a personal-dev token on an encrypted laptop with FileVault; reconsider for shared or production deployments.

The MCP introspects scopes via health_check — call it after wiring up to confirm what your token actually has.

Optional: GITHUB_API_BASE for GitHub Enterprise Server. Defaults to https://api.github.com.

Wire into an MCP client

Claude Code

Add to ~/.claude/.mcp.json (or any .mcp.json in the project root):

{

"mcpServers": { "github": { "command": "uv", "args": ["run", "--project", "/Users/ /.claude/github-mcp", "github-mcp"], "env": { "GITHUB_TOKEN": "${GITHUB_TOKEN}" } } } }

Codex CLI

Add to ~/.codex/config.toml:

[mcp_servers.github]

command = "uv" args = ["run", "--project", "/Users/ /.claude/github-mcp", "github-mcp"]

[mcp_servers.github.env] GITHUB_TOKEN = "${GITHUB_TOKEN}"

Cursor

~/.cursor/mcp.json follows the Claude Code shape. Same env block.

Any other MCP-compliant client

Configure it to spawn the github-mcp command and pipe stdio. The server speaks standard MCP over stdio.

Verify

GITHUB_TOKEN=ghp_yourpat uv run github-mcp

The server will wait on stdin for MCP traffic. In another shell, exercise health_check via your client.

Test

uv pip install -e ".[dev]"

uv run pytest tests/ -v

Telemetry

This plugin sends a single anonymous install signal to myceliumai.co the first time it loads in a Claude Code session on a given machine.

What is sent:

  • Plugin name (e.g. slack-mcp)
  • Plugin version (e.g. 0.1.0)

What is NOT sent:

  • No user identifiers, names, emails, tokens, or API keys
  • No file paths, message content, or anything from your work
  • No IP address is stored after dedup processing

Why: Helps the maintainer know which plugins people actually install, so attention goes to the ones that get used.

Opt out: Set the environment variable MYCELIUM_NO_PING=1 before launching Claude Code. The hook will skip the network call entirely. Already-pinged installs leave a sentinel at ~/.mycelium/onboarded- — delete it if you want to reset state.

License

MIT.

Ask PopNx