Repository-specific endpoints reduce the chance of an agent querying the wrong project, while the dynamic endpoint relies on correct target selection., Verify generated code against the upstream project documentation and changelog before applying it to production systems., Use caution with private, unreleased, or embargoed repositories unless you self-host and have reviewed access controls.
Privacy notes
GitMCP may receive repository choices, documentation queries, tool calls, and surrounding prompts through the MCP client., The README states that the hosted service does not collect personal information or store queries, but teams should still review the current privacy posture before sending proprietary context., Self-hosting may be preferable for private repositories, regulated work, or sensitive customer projects.
Author
idosal
Submitted by
oktofeesh1
Claim status
unclaimed
Last verified
2026-06-05
Decision playbook
Review trust signals before you adopt
Signals are present but mixed. Use the checklist below to confirm the source and operational safety for your environment.
Compare context
Selected
0
Current score
63
Baseline
—
Delta
No baseline selected
No major trust-signal divergence detected in the current selection.
Source and provenance checks
Needs review
Confirm ownership and provenance before trusting install instructions.
Source link availableRequired
Open the canonical repository and verify ownership.
Done
Source provenance statusRequired
Marked as source-backed.
Done
Metadata reviewed
No reviewed flag detected in metadata.
Pending
Safety and privacy checks
Complete
Validate risk disclosures before installation or API wiring.
Safety notes presentRequired
Review the listed safety guidance before running commands.
Done
Privacy notes presentRequired
Review data handling notes before connecting accounts or secrets.
Done
Trust level risk gateRequired
Trust level does not block evaluation.
Done
Package and install checks
Needs review
Check package metadata and artifact integrity signals.
Install payload available
Install or copy payload is available for review.
Done
Package verification flag
No package verification flag provided.
Pending
Checksum metadata
No checksum provided for downloaded artifact.
Pending
Compare-driven decision checks
Needs review
Use compare context to validate trade-offs before adoption.
Compare tray has multiple entries
Add at least one more entry to compare trust differences.
3 safety and 3 privacy notes across 4 risk areas. Review closely: permissions & scopes, network access.
4 areas
SafetyNetwork accessRepository-specific endpoints reduce the chance of an agent querying the wrong project, while the dynamic endpoint relies on correct target selection.
SafetyData retentionVerify generated code against the upstream project documentation and changelog before applying it to production systems.
SafetyPermissions & scopesUse caution with private, unreleased, or embargoed repositories unless you self-host and have reviewed access controls.
PrivacyGeneralGitMCP may receive repository choices, documentation queries, tool calls, and surrounding prompts through the MCP client.
PrivacyData retentionThe README states that the hosted service does not collect personal information or store queries, but teams should still review the current privacy posture before sending proprietary context.
PrivacyGeneralSelf-hosting may be preferable for private repositories, regulated work, or sensitive customer projects.
Safety notes
Repository-specific endpoints reduce the chance of an agent querying the wrong project, while the dynamic endpoint relies on correct target selection.
Verify generated code against the upstream project documentation and changelog before applying it to production systems.
Use caution with private, unreleased, or embargoed repositories unless you self-host and have reviewed access controls.
Privacy notes
GitMCP may receive repository choices, documentation queries, tool calls, and surrounding prompts through the MCP client.
The README states that the hosted service does not collect personal information or store queries, but teams should still review the current privacy posture before sending proprietary context.
Self-hosting may be preferable for private repositories, regulated work, or sensitive customer projects.
Prerequisites
MCP client that supports remote MCP servers or an mcp-remote bridge.
GitHub repository owner and repository name, GitHub Pages site, or the dynamic GitMCP docs endpoint.
Network access to GitMCP and GitHub.
Team agreement on which repositories can be queried through a hosted MCP service.
GitMCP is a hosted, open-source MCP server that turns GitHub repositories and
GitHub Pages sites into documentation and code-context sources for AI coding
assistants. Instead of installing a project-specific MCP server, users point an
MCP client at a GitMCP endpoint for the repository they are working with.
The project supports repository-specific endpoints, GitHub Pages endpoints, and
a dynamic docs endpoint. Repository-specific endpoints are best when an assistant
should stay focused on one project; the dynamic endpoint is useful when the
assistant needs to choose among projects during a session.
These sources were reviewed on 2026-06-05. Prefer the live repository and
GitMCP site for current endpoint formats, supported clients, privacy language,
self-hosting details, and remote MCP compatibility.
Features
Remote MCP access to live GitHub repository documentation and code context.
Repository-specific, GitHub Pages, and dynamic docs endpoint modes.
Setup examples for Cursor, Claude Desktop, Windsurf, VS Code, Cline,
Highlight AI, Augment Code, and Msty.
Smart search over documentation and code sources.
Hosted zero-setup mode plus open-source self-hosting option.
PulseMCP top-pick listing as an external popularity signal.
Installation
For Claude Desktop-style clients that need a bridge, configure mcp-remote:
For clients with native remote server support, add the repository-specific
GitMCP endpoint directly. Replace {owner} and {repo} with the target GitHub
project, or use the dynamic docs endpoint documented in the README.
Use Cases
Ground coding prompts in a library's current README, docs, and examples.
Ask an assistant to search a GitHub project before writing integration code.
Reduce hallucinated APIs for niche or rapidly changing packages.
Give Cursor, Claude Desktop, VS Code, or Windsurf repository-aware docs access
without writing a custom MCP server.
Self-host a GitMCP-compatible service for private documentation workflows.
Safety and Privacy
Prefer repository-specific endpoints when a project should remain scoped to one
codebase. Review all generated changes against the original documentation before
committing them, especially for security-sensitive dependencies or production
infrastructure.
Hosted remote MCP services can see requests routed through them. For private
repositories, customer projects, unreleased APIs, or regulated work, review the
current privacy language and consider self-hosting.
Duplicate Check
content/mcp/git-mcp-server.mdx covers the reference Git MCP server for local
repository operations. This entry covers the separate idosal/git-mcp hosted
documentation and code-context service for GitHub projects.
Show that GitMCP Docs Server is listed on HeyClaude. Paste this Markdown into your README — it renders the badge and links back to this page.
[](https://heyclau.de/entry/mcp/gitmcp-docs-server)
How it compares
GitMCP Docs Server side by side with 3 alternatives on trust, install, platform support, and disclosed safety notes — all from reviewed registry metadata.
1 trust signal differ across this comparison (Submitter).
Comprehensive Google Workspace MCP server and CLI for Gmail, Drive, Calendar, Docs, Sheets, Slides, Forms, Chat, Apps Script, Tasks, Contacts, and Custom Search workflows.
Official AWS Labs fully-managed remote MCP server providing up-to-date AWS documentation, code samples, agent skills, API and CloudFormation regional availability, and CDK/Amplify/Strands guidance over a hosted HTTP endpoint.
Official Microsoft Learn remote MCP server that gives AI agents real-time access to Microsoft documentation search, page fetch, and code sample search.
✓Repository-specific endpoints reduce the chance of an agent querying the wrong project, while the dynamic endpoint relies on correct target selection.
Verify generated code against the upstream project documentation and changelog before applying it to production systems.
Use caution with private, unreleased, or embargoed repositories unless you self-host and have reviewed access controls.
✓Google Workspace MCP can read, create, edit, send, share, delete, and manage data across Gmail, Drive, Calendar, Docs, Sheets, Slides, Forms, Chat, Apps Script, Tasks, Contacts, and Search depending on configured scopes and tool tier.
Start with the smallest useful tool tier and read-only or service-specific scopes where possible.
Require human approval before sending email or chat messages, sharing Drive files, changing permissions, deleting content, editing business documents, or executing Apps Script workflows.
Protect OAuth client secrets, refresh tokens, service account keys, and credential stores.
✓This is a fully managed remote server hosted by AWS; it serves public AWS knowledge and requires no AWS account or credentials, so do not add AWS access keys for it.
Because it is a remote server, your queries are sent to the AWS-hosted endpoint; only connect clients you trust to route requests there.
Generative output about AWS APIs and resources can still be wrong; verify guidance against the official AWS documentation it links.
✓The server is documentation-focused, but retrieved docs can still influence generated commands; review commands before running them.
Remote MCP availability and schema can change, so verify client compatibility before depending on it in automation.
Privacy notes
✓GitMCP may receive repository choices, documentation queries, tool calls, and surrounding prompts through the MCP client.
The README states that the hosted service does not collect personal information or store queries, but teams should still review the current privacy posture before sending proprietary context.
Self-hosting may be preferable for private repositories, regulated work, or sensitive customer projects.
✓Emails, attachments, Drive files, calendar events, documents, spreadsheets, presentations, form responses, chat messages, contacts, search queries, OAuth tokens, and tool logs may be exposed to the MCP client and model.
The repository describes a self-hosted data path to Google APIs, but prompts and tool results still flow through the chosen AI client and model provider.
Review organization data retention, Workspace admin policies, OAuth app verification, and model-provider retention before connecting production accounts.
✓Your documentation queries and search terms are sent over HTTPS to the AWS-hosted knowledge endpoint to retrieve content.
The server returns public AWS documentation, code samples, and availability data; it does not read any private account data.
✓Prompts and queries sent to a remote MCP server can reveal product names, architecture details, or error messages.
Avoid sending customer identifiers or private tenant data when documentation search is enough.
Prerequisites
MCP client that supports remote MCP servers or an mcp-remote bridge.
GitHub repository owner and repository name, GitHub Pages site, or the dynamic GitMCP docs endpoint.
Network access to GitMCP and GitHub.
Team agreement on which repositories can be queried through a hosted MCP service.
Python 3.10 or newer and uv or uvx.
Google Cloud project with OAuth credentials.
Required Google Workspace APIs enabled for the services you plan to use.
OAuth consent, scopes, and redirect settings reviewed for the target account or organization.
An MCP client that supports remote HTTP MCP servers (or `uvx`/`fastmcp` to proxy the endpoint to stdio for clients that don't).
Outbound HTTPS access to the hosted endpoint `https://knowledge-mcp.global.api.aws`.
An MCP-compatible client that supports remote HTTP MCP servers.
Network access to learn.microsoft.com.
A Microsoft technology question where official Learn grounding is useful.
Install
npx mcp-remote https://gitmcp.io/{owner}/{repo}
uvx workspace-mcp --tool-tier core
uvx fastmcp run https://knowledge-mcp.global.api.aws
claude mcp add --transport http microsoft-learn https://learn.microsoft.com/api/mcp