20 Best GitHub MCP Servers, Compared (September 2026)
The short answer
Use the GitHub MCP Server (github/github-mcp-server) unless your task is narrow. It is GitHub's official MCP server, is labelled Community favourite, and is active: it had a commit 6 days ago and 200 commits in the last 12 weeks. That covers the common needs of repository browsing, issue and PR automation, and Actions workflow monitoring. If your task is strictly to pull specific functions or classes out of a large repository while keeping token use low, github-codemunch-mcp is the better answer: it had a commit 0 days ago and 1,010 commits in the last 12 weeks.
Whichever you choose, give it the narrowest access that still works (a read-only credential, a replica, a scratch account), and widen it only once you have watched what your agent actually asks for.
Glama operates the MCP registry these numbers are measured from, and sells MCP hosting and a gateway. No position on this page is paid for. How the registry is built.
Quick picks
- 1GitHub MCP Server : Teams that want GitHub's own MCP server for AI assistants: it documents repository browsing, issue and PR automation, and Actions workflow monitoring.`n`n`
- 2github-codemunch-mcp : For exploring large GitHub repositories and retrieving only the specific functions or classes you need, its tree-sitter AST indexing enables precise symbol search and extraction.
- 3Octocode MCP : Developers researching unfamiliar codebases across local files, GitHub, and npm: it combines local ripgrep/LSP tools with GitHub and package search.
- 4idea-reality-mcp : Before starting a new project: it exposes idea_check, which scans GitHub, Hacker News, npm, PyPI and Stack Overflow for existing competition.
- 5mcp-github-project-manager : AI coding agents that need a persistent coordination layer on GitHub Projects v2: it provides task checkout, atomic claims, heartbeat monitoring, and token budgets.
Which one, for your situation
| Your situation | What to use |
|---|---|
| I need an official maintained server for repository and PR work | Use the GitHub MCP Server (github/github-mcp-server): it is GitHub's official server, had a commit 6 days ago, and has 200 commits in the last 12 weeks. |
| Large repo, I need only specific functions or classes with low tokens | Use github-codemunch-mcp: it uses tree-sitter AST for symbol extraction, exposes 7 tools, and had a commit 0 days ago and 1,010 commits in the last 12 weeks. |
| I manage GitHub Projects v2 boards, sprints, and tasks | Use mcp-github-project-manager: it is built for GitHub Projects v2 workflows and exposes 84 tools. |
| I need a one-command GitHub org and repository security audit | Use github-security-mcp: it has 39 tools for GitHub security audits, but it is labelled Abandoned but popular, with a last commit 184 days ago and 0 commits in the last 12 weeks. |
| Detect duplicate code before I commit a snippet | Use jscpd: its MCP server can check a snippet against a scanned project, exposes 4 tools, and had a commit 0 days ago and 370 commits in the last 12 weeks. |
| Wrap my existing MCP server in a security proxy | Use pipelock: it wraps any MCP server with bidirectional scanning for credential leaks, prompt injection, and tool description poisoning, and had a commit 0 days ago and 700 commits in the last 12 weeks. |
Top MCP servers for GitHub
| Best for | Profile | ||||||
|---|---|---|---|---|---|---|---|
| 1 | Teams that want GitHub's own MCP server for AI assistants: it documents repository browsing, issue and PR automation, and Actions workflow monitoring.`n`n` | Community favourite | 32,846 | +680 | 6 days ago | 92.3 | |
| 2 | For exploring large GitHub repositories and retrieving only the specific functions or classes you need, its tree-sitter AST indexing enables precise symbol search and extraction. | Community favourite | 2,684 | +123 | today | 82.8 | |
| 3 | Developers researching unfamiliar codebases across local files, GitHub, and npm: it combines local ripgrep/LSP tools with GitHub and package search. | Community favourite | 932 | +21 | 20 days ago | 66.5 | |
| 4 | Before starting a new project: it exposes idea_check, which scans GitHub, Hacker News, npm, PyPI and Stack Overflow for existing competition. | Community favourite | 815 | +36 | 34 days ago | 64.7 | |
| 5 | AI coding agents that need a persistent coordination layer on GitHub Projects v2: it provides task checkout, atomic claims, heartbeat monitoring, and token budgets. | Steady | 97 | +4 | 6 days ago | 63.6 | |
| 6 | Debugging a coding problem that needs StackOverflow, GitHub Issues, arXiv, or Wikipedia content: it returns the full conversation in one call. | Community favourite | 385 | +14 | 3 days ago | 59.2 | |
| 7 | Developers using AI assistants like Cursor with niche or recently changed GitHub projects: it exposes up-to-date repository docs and code through a URL-based MCP endpoint. | Abandoned but popular | 8,376 | +57 | 129 days ago | 55.7 | |
| 8 | When you need to convert docs, GitHub repos, PDFs, videos, and codebases into AI-ready skills from Claude Code, it exposes scraping, packaging, and upload tools. | Community favourite | 14,963 | +208 | yesterday | 54.7 | |
| 9 | Incident response and dependency audits: it exposes 41 tools that pull CVE, exploit, and package-impact data from 11 vulnerability sources through one MCP server. | Community favourite | 27 | +8 | 70 days ago | 54.4 | |
| 10 | Finding a previously starred repository from a description: it indexes the READMEs of a user's starred repositories and provides natural-language search. | Steady | 113 | +1 | 2 days ago | 54.1 | |
| 11 | AI agents that need a one-command GitHub org and repo security audit: it exposes a tool that runs all 45 checks and aggregates findings. | Abandoned but popular | 12 | 0 | 184 days ago | 52.6 | |
| 12 | Before writing or committing a block: check_duplication tests a snippet against the scanned project. | Community favourite | 6,193 | No snapshot history | today | 50.4 | |
| 13 | Wrapping an existing MCP server in a security boundary: it scans both directions for credential leaks, prompt injection, and tool description poisoning. | Community favourite | 842 | +51 | today | 49.7 | |
| 14 | From a written spec to a verified product build: it runs autonomously and exposes MCP tools for task queue, code search, and verification reports. | Community favourite | 1,057 | +16 | today | 48.8 | |
| 15 | Pre-merge review of AI-generated code: it exposes static-analysis tools such as security_scan and validate_code_change for dead code, secrets, and quality issues. | Community favourite | 746 | +291 | today | 47.2 | |
| 16 | For governing an autonomous AI agent's tool use: Node9-Proxy sits between agent and tools to discover, protect, and review actions. | Community favourite | 211 | +2 | today | 46.6 | |
| 17 | Runaway agent loops on Claude Code or Cursor: it enforces per-session spend caps, verifier safety gates, and runaway loop prevention. | Emerging | 51 | +9 | today | 46.0 | |
| 18 | Managing GitHub repositories and files: it provides tools for file operations, repository management, and search. | Abandoned but popular | 90,341 | +726 | 419 days ago | 45.4 | |
| 19 | GitHub code, repository, or user search plus web search, AI answers, and extraction via an MCP server: it exposes a single interface for all. | Community favourite | 347 | +5 | today | 44.0 | |
| 20 | For organizations wanting GitHub discussions, issues, and pull request summaries beyond the 100-item API limit: this server stores GitHub data locally and answers in Markdown. | Dormant | 78 | 0 | 222 days ago | 43.8 |
The ranking, with the evidence
Each position is a weighted mean of adoption (40%), maintenance (24%), momentum (14%), tool description quality (13%) and trust (9%), multiplied by three attenuators: how directly the server is about GitHub (named for it, declaring it, tagged with it, or merely mentioning it), whether its repository is still moving, and how much independent evidence of adoption it has. Open the score on any entry to see every number, including the ones marked ≈, which were imputed from the median of the other candidates rather than measured. The maintenance grade on each entry is mostly issue responsiveness, release recency and open security alerts rather than commits, so a recent commit beside a low grade is two different measurements rather than a contradiction.
- Abandoned but popular: People use it, but its default branch has stopped moving. Fine to keep running, risky to adopt.
- Community favourite: Widely adopted and still actively maintained.
- Dormant: Neither changing nor widely adopted. Here because it still matches the search.
- Emerging: Small audience, growing quickly, maintained. The bet with the most upside.
- Steady: Maintained, modest audience, no surprises in either direction.
Best for: Teams that want GitHub's own MCP server for AI assistants: it documents repository browsing, issue and PR automation, and Actions workflow monitoring.`n`n`
The README documents an MCP server that connects AI tools to GitHub for natural language work with repositories, issues, pull requests, Actions, and security findings. Before choosing it, note that the remote server is hosted by GitHub and requires a compatible MCP host with remote server support, although a local version exists.
GitHub stars32,846Stars / 30 days+680npm / typical weekShips no npm packagePyPI / typical weekno attributed packageTools exposednever inspectedLast commit6 days agoCommits / 12 weeks200Maintenance gradeATool descriptionsNot gradedScore 92.3: show every number behind it
- Adoption100 / 100 · weight 40%
- GitHub stars100
- npm downloadsnot measuredno npm package
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum75 / 100 · weight 14%
- Stars gained, relative to size59
- Stars gained, absolute100
- npm download trendnot measuredno download history for the selected registry
- Tool quality≈68 / 100 · weight 13%
- Tool description quality≈68tool descriptions not yet scored
- Built and inspected by Glamanot measurednever built and inspected by Glama
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integrates100
- Weighted mean of the five
- 92.3
- × relevance: the keyword is dedicated here
- 1.00
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 92.3
Best for: For exploring large GitHub repositories and retrieving only the specific functions or classes you need, its tree-sitter AST indexing enables precise symbol search and extraction.
It indexes a repository locally, then exposes tools to list files, outline symbols, and fetch full source for matched functions or classes. Before choosing it, note that it requires building a local index first, and it is distributed as a Python package rather than an npm module.
GitHub stars2,684Stars / 30 days+123npm / typical weekShips no npm packagePyPI / typical week25Kjcodemunch-mcpTools exposed7Last committodayCommits / 12 weeks1,010Maintenance gradeATool descriptionsAScore 82.8: show every number behind it
- Adoption100 / 100 · weight 40%
- GitHub stars86
- PyPI downloads94
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum64 / 100 · weight 14%
- Stars gained, relative to size86
- Stars gained, absolute84
- PyPI download trend0
- Tool quality76 / 100 · weight 13%
- Tool description quality68
- Built and inspected by Glama100
- Trust0 / 100 · weight 9%
- License0
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 82.8
- × relevance: the keyword is dedicated here
- 1.00
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 82.8
Best for: Developers researching unfamiliar codebases across local files, GitHub, and npm: it combines local ripgrep/LSP tools with GitHub and package search.
Octocode exposes 13 tools that cover local code search, file reading, LSP navigation, GitHub repository and PR search, and npm/PyPI package lookup. It requires Node.js 20.12+ and runs via npx, with GitHub authentication optional but recommended for private repositories and higher rate limits.
GitHub stars932Stars / 30 days+21npm / typical week2.8KPyPI / typical weekno attributed packageTools exposed13Last commit20 days agoCommits / 12 weeks24Maintenance gradeATool descriptionsAScore 66.5: show every number behind it
- Adoption85 / 100 · weight 40%
- GitHub stars74
- npm downloads73
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance97 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence85
- Momentum42 / 100 · weight 14%
- Stars gained, relative to size58
- Stars gained, absolute54
- npm download trend0
- Tool quality100 / 100 · weight 13%
- Tool description quality93
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 85.3
- × relevance: the keyword is declared here
- 0.78
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 66.5
Best for: Before starting a new project: it exposes idea_check, which scans GitHub, Hacker News, npm, PyPI and Stack Overflow for existing competition.
It exposes one tool, idea_check, which takes a plain-English idea description and returns a reality signal with trend direction, top competitors and pivot suggestions. It is installed with uvx and needs no API key; the README also documents a REST API.
GitHub stars815Stars / 30 days+36npm / typical weekShips no npm packagePyPI / typical week84idea-reality-mcpTools exposed1Last commit34 days agoCommits / 12 weeks36Maintenance gradeBTool descriptionsAScore 64.7: show every number behind it
- Adoption76 / 100 · weight 40%
- GitHub stars73
- PyPI downloads21PyPI downloads show no weekday rhythm; halved
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance95 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade80
- Commit cadence100
- Momentum62 / 100 · weight 14%
- Stars gained, relative to size81
- Stars gained, absolute63
- PyPI download trend26
- Tool quality93 / 100 · weight 13%
- Tool description quality85
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 82.9
- × relevance: the keyword is declared here
- 0.78
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 64.7
Best for: AI coding agents that need a persistent coordination layer on GitHub Projects v2: it provides task checkout, atomic claims, heartbeat monitoring, and token budgets.
The server exposes tools for roadmaps, sprint planning, milestones, issues, draft issues, pull requests, and GitHub projects, and the README documents agent self-assignment, heartbeat monitoring, review workflows, and token budgets. Choosing it requires GITHUB_TOKEN, GITHUB_OWNER, and GITHUB_REPO, and assumes all coordination state lives in GitHub issues, projects, and comments.
GitHub stars97Stars / 30 days+4npm / typical week49PyPI / typical weekno attributed packageTools exposed84Last commit6 days agoCommits / 12 weeks68Maintenance gradeCTool descriptionsCScore 63.6: show every number behind it
- Adoption55 / 100 · weight 40%
- GitHub stars50
- npm downloads36
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance89 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade55
- Commit cadence100
- Momentum33 / 100 · weight 14%
- Stars gained, relative to size54
- Stars gained, absolute29
- npm download trend0
- Tool quality51 / 100 · weight 13%
- Tool description quality43
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 63.6
- × relevance: the keyword is dedicated here
- 1.00
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 63.6
- 6
Best for: Debugging a coding problem that needs StackOverflow, GitHub Issues, arXiv, or Wikipedia content: it returns the full conversation in one call.
The README documents web_search and get_content, which return search results and page content in Markdown, and says it integrates with StackOverflow, GitHub Issues, arXiv, and Wikipedia. Before choosing it, note that it has no published npm package and no attributed PyPI package.
GitHub stars385Stars / 30 days+14npm / typical weekShips no npm packagePyPI / typical weekno attributed packageTools exposednever inspectedLast commit3 days agoCommits / 12 weeks278Maintenance gradeATool descriptionsNot gradedScore 59.2: show every number behind it
- Adoption65 / 100 · weight 40%
- GitHub stars65
- npm downloadsnot measuredno npm package
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum59 / 100 · weight 14%
- Stars gained, relative to size67
- Stars gained, absolute48
- npm download trendnot measuredno download history for the selected registry
- Tool quality≈68 / 100 · weight 13%
- Tool description quality≈68tool descriptions not yet scored
- Built and inspected by Glamanot measurednever built and inspected by Glama
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 75.9
- × relevance: the keyword is declared here
- 0.78
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 59.2
Best for: Developers using AI assistants like Cursor with niche or recently changed GitHub projects: it exposes up-to-date repository docs and code through a URL-based MCP endpoint.
It exposes 32 tools spanning Git operations, GitHub and Gitea repository management, releases, auth, configuration, backups, and health checks. It is a hosted cloud MCP endpoint reached by URL, so clients need network access to gitmcp.io unless the server is self-hosted.
GitHub stars8,376Stars / 30 days+57npm / typical weekdownloads not countedPyPI / typical weekno attributed packageTools exposed32Last commit129 days agoCommits / 12 weeks0Maintenance gradeFTool descriptionsCScore 55.7: show every number behind it
- Adoption98 / 100 · weight 40%
- GitHub stars98
- npm downloadsnot measurednpm names no repository for git-mcp, so its downloads cannot be attributed
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance40 / 100 · weight 24%
- Last commit touching this server68
- Repository maintenance grade5
- Commit cadence5
- Momentum49 / 100 · weight 14%
- Stars gained, relative to size34
- Stars gained, absolute71
- npm download trendnot measuredno download history for the selected registry
- Tool quality53 / 100 · weight 13%
- Tool description quality45
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 71.4
- × relevance: the keyword is declared here
- 0.78
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 55.7
Best for: When you need to convert docs, GitHub repos, PDFs, videos, and codebases into AI-ready skills from Claude Code, it exposes scraping, packaging, and upload tools.
The server exposes 40 tools for scraping documentation, GitHub repositories, PDFs, videos, and local codebases, generating configs, packaging skills, uploading to LLM platforms, exporting to vector databases, and managing workflows and marketplaces. It assumes the Python package is installed with the mcp extra, because the core install does not include the MCP server.
GitHub stars14,963Stars / 30 days+208npm / typical weekShips no npm packagePyPI / typical week3.4Kskill-seekersTools exposed40Last commityesterdayCommits / 12 weeks66Maintenance gradeATool descriptionsBScore 54.7: show every number behind it
- Adoption100 / 100 · weight 40%
- GitHub stars100
- PyPI downloads38PyPI downloads show no weekday rhythm; halved
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum74 / 100 · weight 14%
- Stars gained, relative to size48
- Stars gained, absolute94
- PyPI download trend96
- Tool quality61 / 100 · weight 13%
- Tool description quality53
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 91.2
- × relevance: the keyword is tagged here
- 0.60
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 54.7
Best for: Incident response and dependency audits: it exposes 41 tools that pull CVE, exploit, and package-impact data from 11 vulnerability sources through one MCP server.
The server exposes 41 MCP tools for querying NVD, EPSS, CISA KEV, GitHub Advisory, OSV, Shodan, VulnCheck, Vulners, Nuclei, Metasploit, CIRCL, and AttackerKB, including CVE enrichment, exploit checks, and risk scoring. Before installing it, know that it is published as an npm package and the README specifies the Bun runtime.
GitHub stars27Stars / 30 days+8npm / typical week413PyPI / typical weekno attributed packageTools exposed41Last commit70 days agoCommits / 12 weeks5Maintenance gradeBTool descriptionsAScore 54.4: show every number behind it
- Adoption61 / 100 · weight 40%
- GitHub stars36
- npm downloads56
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance81 / 100 · weight 24%
- Last commit touching this server88dated from the last commit on the default branch, re-read from GitHub at publication; github.com shows a push 5 days ago, which counts every ref; the stored date would have published 5 days ago
- Repository maintenance grade80
- Commit cadence65
- Momentum50 / 100 · weight 14%
- Stars gained, relative to size86
- Stars gained, absolute37
- npm download trend0
- Tool quality76 / 100 · weight 13%
- Tool description quality68
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 69.7
- × relevance: the keyword is declared here
- 0.78
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 54.4
Best for: Finding a previously starred repository from a description: it indexes the READMEs of a user's starred repositories and provides natural-language search.
The README documents a search_github_stars tool that takes a natural-language query and returns matching starred repositories with relevant README content. Choosing it assumes a Cloudflare-backed deployment: a Cloudflare account, an R2 bucket, AutoRAG configuration, and a GitHub token with repo scope are all required.
GitHub stars113Stars / 30 days+1npm / typical weekdownloads not countedPyPI / typical weekno attributed packageTools exposednever inspectedLast commit2 days agoCommits / 12 weeks15Maintenance gradeBTool descriptionsNot gradedScore 54.1: show every number behind it
- Adoption51 / 100 · weight 40%
- GitHub stars51
- npm downloadsnot measurednpm names no repository for mcp-github-stars, so its downloads cannot be attributed
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance92 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade80
- Commit cadence85
- Momentum23 / 100 · weight 14%
- Stars gained, relative to size29
- Stars gained, absolute15
- npm download trendnot measuredno download history for the selected registry
- Tool quality≈68 / 100 · weight 13%
- Tool description quality≈68tool descriptions not yet scored
- Built and inspected by Glamanot measurednever built and inspected by Glama
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 63.6
- × relevance: the keyword is dedicated here
- 1.00
- × continuity: actively changing
- 1.00
- × evidence: modest but real audience
- 0.85
- Composite score
- 54.1
Best for: AI agents that need a one-command GitHub org and repo security audit: it exposes a tool that runs all 45 checks and aggregates findings.
It exposes 39 tools that audit organization settings, repositories, Actions workflows, secrets, supply chain, and access control, and a single tool can execute all checks sequentially. It runs under the Bun runtime and expects a GitHub personal access token with admin:org and repo scopes.
GitHub stars12Stars / 30 days0npm / typical week298PyPI / typical weekno attributed packageTools exposed39Last commit184 days agoCommits / 12 weeks0Maintenance gradeDTool descriptionsAScore 52.6: show every number behind it
- Adoption57 / 100 · weight 40%
- GitHub stars28
- npm downloads53
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance46 / 100 · weight 24%
- Last commit touching this server68
- Repository maintenance grade30
- Commit cadence5
- Momentum0 / 100 · weight 14%
- Stars gained, relative to size0
- Stars gained, absolute0
- npm download trend0
- Tool quality76 / 100 · weight 13%
- Tool description quality68
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 52.6
- × relevance: the keyword is dedicated here
- 1.00
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 52.6
Best for: Before writing or committing a block: check_duplication tests a snippet against the scanned project.
It exposes four tools: check_duplication for snippets, get_file_clones for file clones, get_statistics for scan statistics, and check_current_directory to rescan. The scope to know: duplication checks apply only to paths the server was started with, so re-scanning via check_current_directory is needed after edits.
GitHub stars6,193Stars / 30 daysno snapshot historynpm / typical weekShips no npm packagePyPI / typical weekno attributed packageTools exposed4Last committodayCommits / 12 weeks370Maintenance gradeATool descriptionsAScore 50.4: show every number behind it
- Adoption95 / 100 · weight 40%
- GitHub stars95
- npm downloadsnot measuredno npm package
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum≈0 / 100 · weight 14%
- Stars gained, relative to sizenot measuredno snapshot history yet
- Stars gained, absolutenot measuredno snapshot history yet
- npm download trendnot measuredno download history for the selected registry
- Tool quality100 / 100 · weight 13%
- Tool description quality93
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integrates100
- Weighted mean of the five
- 84.0
- × relevance: the keyword is tagged here
- 0.60
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 50.4
Best for: Wrapping an existing MCP server in a security boundary: it scans both directions for credential leaks, prompt injection, and tool description poisoning.
It is a proxy that sits between AI agents and the network, inspecting mediated HTTP, WebSocket, MCP, and A2A traffic, plus CONNECT tunnel contents when TLS interception is enabled, for secret exfiltration, prompt injection, SSRF, tool poisoning, and risky tool-call chains. Plain CONNECT tunnels without TLS interception are scanned only at the hostname and URL level: enabling TLS interception is required to scan their contents.
GitHub stars842Stars / 30 days+51npm / typical weekShips no npm packagePyPI / typical weekno attributed packageTools exposednever inspectedLast committodayCommits / 12 weeks700Maintenance gradeATool descriptionsNot gradedScore 49.7: show every number behind it
- Adoption73 / 100 · weight 40%
- GitHub stars73
- npm downloadsnot measuredno npm package
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum85 / 100 · weight 14%
- Stars gained, relative to size95
- Stars gained, absolute69
- npm download trendnot measuredno download history for the selected registry
- Tool quality≈68 / 100 · weight 13%
- Tool description quality≈68tool descriptions not yet scored
- Built and inspected by Glamanot measurednever built and inspected by Glama
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 82.8
- × relevance: the keyword is tagged here
- 0.60
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 49.7
Best for: From a written spec to a verified product build: it runs autonomously and exposes MCP tools for task queue, code search, and verification reports.
The server is the MCP interface to Loki Mode, exposing stdio tools for project state and task-queue operations, memory retrieve/store, code search, quality and verification reports, repo hotspots/co-changes, and structured findings/learnings. The server has never been inspected, so its tool list is unknown.
GitHub stars1,057Stars / 30 days+16npm / typical week2.7KPyPI / typical weekno attributed packageTools exposednever inspectedLast committodayCommits / 12 weeks956Maintenance gradeATool descriptionsNot gradedScore 48.8: show every number behind it
- Adoption87 / 100 · weight 40%
- GitHub stars76
- npm downloads73
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum36 / 100 · weight 14%
- Stars gained, relative to size47
- Stars gained, absolute49
- npm download trend0
- Tool quality≈68 / 100 · weight 13%
- Tool description quality≈68tool descriptions not yet scored
- Built and inspected by Glamanot measurednever built and inspected by Glama
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 81.3
- × relevance: the keyword is tagged here
- 0.60
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 48.8
Best for: Pre-merge review of AI-generated code: it exposes static-analysis tools such as security_scan and validate_code_change for dead code, secrets, and quality issues.
The server exposes 12 tools for static analysis, including analyze, security_scan, secrets_scan, quality_check, remediate, and validate_code_change for checking a diff before it lands. It runs locally by default and can also generate a GitHub Actions PR gate with
skylos cicd init.GitHub stars746Stars / 30 days+291npm / typical weekShips no npm packagePyPI / typical week19.6KskylosTools exposed12Last committodayCommits / 12 weeks156Maintenance gradeATool descriptionsDScore 47.2: show every number behind it
- Adoption79 / 100 · weight 40%
- GitHub stars72
- PyPI downloads46PyPI downloads show no weekday rhythm; halved
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum75 / 100 · weight 14%
- Stars gained, relative to size100
- Stars gained, absolute99
- PyPI download trend0
- Tool quality28 / 100 · weight 13%
- Tool description quality20
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 78.7
- × relevance: the keyword is tagged here
- 0.60
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 47.2
Best for: For governing an autonomous AI agent's tool use: Node9-Proxy sits between agent and tools to discover, protect, and review actions.
Node9-Proxy ships as the npm package @node9/proxy and, per its README, interposes between an AI agent and its tools with CLI commands for scanning sessions, checking posture, scanning repos, sandboxing agents, and enabling shields. Before choosing it, note that its MCP tool list is unknown: this server has never been inspected.
GitHub stars211Stars / 30 days+2npm / typical week1.2KPyPI / typical weekno attributed packageTools exposednever inspectedLast committodayCommits / 12 weeks621Maintenance gradeATool descriptionsNot gradedScore 46.6: show every number behind it
- Adoption74 / 100 · weight 40%
- GitHub stars58
- npm downloads65
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum44 / 100 · weight 14%
- Stars gained, relative to size30
- Stars gained, absolute19
- npm download trend100
- Tool quality≈68 / 100 · weight 13%
- Tool description quality≈68tool descriptions not yet scored
- Built and inspected by Glamanot measurednever built and inspected by Glama
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 77.7
- × relevance: the keyword is tagged here
- 0.60
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 46.6
Best for: Runaway agent loops on Claude Code or Cursor: it enforces per-session spend caps, verifier safety gates, and runaway loop prevention.
MartinLoop exposes 24 MCP tools that govern an AI coding run: martin_run executes a run after workflow admission, while martin_plan, martin_preflight, martin_pause, and martin_cancel shape the run before and during execution. It requires a configured verifier before it will mark a run VERIFIED, and that marker only proves the checks that verifier ran, not that the code is bug-free or safe to merge.
GitHub stars51Stars / 30 days+9npm / typical week242PyPI / typical weekno attributed packageTools exposed24Last committodayCommits / 12 weeks437Maintenance gradeATool descriptionsAScore 46.0: show every number behind it
- Adoption57 / 100 · weight 40%
- GitHub stars43
- npm downloads51
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum74 / 100 · weight 14%
- Stars gained, relative to size86
- Stars gained, absolute40
- npm download trend95
- Tool quality81 / 100 · weight 13%
- Tool description quality73
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 76.7
- × relevance: the keyword is tagged here
- 0.60
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 46.0
Best for: Managing GitHub repositories and files: it provides tools for file operations, repository management, and search.
It provides MCP tools for the GitHub API covering file operations, repository management, and search. Its tool list has not been inspected, so specific capabilities must be confirmed from the repository.
GitHub stars90,3416,453 after dividing by the 14 servers in this repositoryStars / 30 days+726npm / typical weekdownloads not countedPyPI / typical weekno attributed packageTools exposednever inspectedLast commit419 days agoCommits / 12 weeks71Maintenance gradeATool descriptionsNot gradedScore 45.4: show every number behind it
- Adoption95 / 100 · weight 40%
- GitHub stars95
- npm downloadsnot measurednpm names no repository for @modelcontextprotocol/server-github, so its downloads cannot be attributed
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance42 / 100 · weight 24%
- Last commit touching this server42dated from the last commit on the default branch, re-read from GitHub at publication; github.com shows a push 12 days ago, which counts every ref
- Repository maintenance gradenot measuredrepository is active but this server inside it is not
- Commit cadencenot measuredrepository is active but this server inside it is not
- Momentum62 / 100 · weight 14%
- Stars gained, relative to size37
- Stars gained, absolute100
- npm download trendnot measuredno download history for the selected registry
- Tool quality≈76 / 100 · weight 13%
- Tool description quality≈68tool descriptions not yet scored
- Built and inspected by Glama100
- Trust≈100 / 100 · weight 9%
- License≈100no license detected
- Published by the vendor it integrates100
- Weighted mean of the five
- 75.7
- × relevance: the keyword is dedicated here
- 1.00
- × continuity: no change in over a year
- 0.60
- × evidence: widely adopted
- 1.00
- Composite score
- 45.4
Best for: GitHub code, repository, or user search plus web search, AI answers, and extraction via an MCP server: it exposes a single interface for all.
It exposes web search, AI search, and web extraction tools; its README also documents a GitHub search tool. It depends on external provider API keys, but providers without a key are skipped and the rest keep working.
GitHub stars347Stars / 30 days+5npm / typical weekdownloads not countedPyPI / typical weekno attributed packageTools exposed3Last committodayCommits / 12 weeks93Maintenance gradeATool descriptionsAScore 44.0: show every number behind it
- Adoption64 / 100 · weight 40%
- GitHub stars64
- npm downloadsnot measurednpm names no repository for mcp-omnisearch, so its downloads cannot be attributed
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance100 / 100 · weight 24%
- Last commit touching this server100
- Repository maintenance grade100
- Commit cadence100
- Momentum39 / 100 · weight 14%
- Stars gained, relative to size43
- Stars gained, absolute33
- npm download trendnot measuredno download history for the selected registry
- Tool quality73 / 100 · weight 13%
- Tool description quality65
- Built and inspected by Glama100
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 73.3
- × relevance: the keyword is tagged here
- 0.60
- × continuity: actively changing
- 1.00
- × evidence: widely adopted
- 1.00
- Composite score
- 44.0
- 20
Best for: For organizations wanting GitHub discussions, issues, and pull request summaries beyond the 100-item API limit: this server stores GitHub data locally and answers in Markdown.
The README documents a stdio MCP server with a TUI for GitHub login, pulling data into a local database, and configuration; the MCP server then responds to summary questions using Markdown output. Before choosing it, note that it requires an ORGANIZATION value and an initial pull before the MCP server can serve summaries.
GitHub stars78Stars / 30 days0npm / typical week14PyPI / typical weekno attributed packageTools exposednever inspectedLast commit222 days agoCommits / 12 weeks0Maintenance gradeBTool descriptionsNot gradedScore 43.8: show every number behind it
- Adoption49 / 100 · weight 40%
- GitHub stars47
- npm downloads13downloads show none of the weekday rhythm human traffic has; halved
- Used through Glamanot measurednot used through Glama in the last 30 days
- Maintenance58 / 100 · weight 24%
- Last commit touching this server68
- Repository maintenance grade80
- Commit cadence5
- Momentum0 / 100 · weight 14%
- Stars gained, relative to size0
- Stars gained, absolute0
- npm download trend0
- Tool quality≈68 / 100 · weight 13%
- Tool description quality≈68tool descriptions not yet scored
- Built and inspected by Glamanot measurednever built and inspected by Glama
- Trust100 / 100 · weight 9%
- License100
- Published by the vendor it integratesnot measurednot published by the vendor it integrates
- Weighted mean of the five
- 51.5
- × relevance: the keyword is dedicated here
- 1.00
- × continuity: actively changing
- 1.00
- × evidence: modest but real audience
- 0.85
- Composite score
- 43.8
Questions people ask
Should I use github/github-mcp-server or modelcontextprotocol/servers?
Use the GitHub MCP Server (github/github-mcp-server). It is GitHub's official server, is labelled Community favourite, and had a commit 6 days ago and 200 commits in the last 12 weeks. The modelcontextprotocol/servers variant is labelled Abandoned but popular and had a last commit 419 days ago.
When would I prefer github-codemunch-mcp over the official one?
Prefer github-codemunch-mcp when you need to retrieve specific functions or classes from a large repository while reducing token use. Its described approach is tree-sitter AST parsing to extract symbol signatures and summaries. It exposes 7 tools and had a commit 0 days ago and 1,010 commits in the last 12 weeks.
Which security server should I pick: github-security-mcp or cve-mcp?
Use github-security-mcp for a GitHub org and repository security audit; it has 39 tools. Be aware that it is labelled Abandoned but popular, with a last commit 184 days ago and 0 commits in the last 12 weeks. Use cve-mcp for vulnerability intelligence from sources including GitHub Advisory and OSV; it has 41 tools and had a commit 70 days ago and 5 commits in the last 12 weeks.
What should I be careful about before using mcp-skylos?
mcp-skylos is actively maintained, with a last commit 0 days ago and 156 commits in the last 12 weeks. It exposes 12 static analysis tools, but its tool description grade is D. Test how the static-analysis tools are understood by your agent before relying on them.
What can guard an agent's actions against leaks and wasted tool calls?
pipelock scans both directions for credential leaks, prompt injection, and tool description poisoning, and it had a commit 0 days ago and 700 commits in the last 12 weeks. Node9-Proxy sits between the agent and tools for discovery, protection, and review, and had a commit 0 days ago and 621 commits in the last 12 weeks. MartinLoop is the one aimed at session spend caps and runaway loops.