Every RepoRadar-tracked MCP server, gateway, registry, and MCP-enabled tool in one view. Same evidence-linked scoring, risk labels, and verdicts as the full catalog. Default sort is score high to low.
Research topics (model-assisted): authoritative editorial categories and high-margin candidate suggestions are included by default. Suggestions never affect score, verdict, badges, comparisons, curated eligibility, or the authoritative category.
G#001GoldTool
awslabs/mcp
What it is: awslabs/mcp is a maintained catalog of production AWS MCP servers with documented setup paths for retrieving and acting on AWS context safely from agent clients.
Why it matters: Useful for teams on AWS who need practical MCP integrations without hand-building service-specific glue.
Score9
Popularity93
RiskConditional
VerificationTested in a bounded workflow
Score breakdown
Usefulness9.0
Novelty8.0
Momentum9.0
Maturity8.9
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
platform engineers standardizing on AWSbuilders adding tool-calling workflowsenterprise teams preferring official AWS integrations
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Mcp Infrastructure · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: Tested in a bounded workflowA retained bounded workflow records setup, action, outcome, limitation, cleanup, and log evidence. Basis: Bounded representative workflow retained by RepoRadar verification harness. Last checked 2026-07-14.
First seen: 2026-06-17T23:02:49Z · Updated: 2026-06-17T23:02:49Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#002GoldTool / Platform
github/github-mcp-server
What it is: GitHub's official MCP server exposes GitHub workflows, repos, issues, and code context through a standards-based tool surface for AI assistants, so tooling can be connected via a single Model Context Protocol endpoint.
Why it matters: Useful for teams already living in GitHub who need stable MCP access to repository data and issue/workflow actions without a custom server layer.
Score9
Popularity86
RiskNone
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum9.0
Maturity8.7
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
engineering teams that already authenticate against GitHubagent builders that need issue/repo/workflow contextteams building custom AI assistants with enterprise GitHub permissions
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Mcp Infrastructure · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
No inherent risk flagged.
Caution flags
not hands-on tested
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-17T17:10:00Z · Updated: 2026-06-17T17:10:00Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#003GoldFramework
microsoft/mcp-gateway
What it is: MCP Gateway is Microsoft's reverse proxy and management layer for MCP servers, adding session-aware routing, auth controls, server lifecycle management, and Kubernetes-friendly deployment paths for teams that need more than a single local MCP process.
Why it matters: Useful for platform teams that are moving from one-off MCP experiments to shared infrastructure: it gives them a cleaner way to route tool traffic, keep sessions sticky, and manage multiple servers without inventing the control plane from scratch.
Score8.9
Popularity78
RiskMedium
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum8.0
Maturity8.5
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease4.2
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
platform teamsdeveloper infrastructure engineersMCP server operators
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Mcp Infrastructure · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Medium risk from workflow or data-surface assumptions in 'MCP Infrastructure / Kubernetes'.
Caution flags
not hands-on testedrestricted permissionsdata handling risk
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-19T04:01:59Z · Updated: 2026-06-19T04:01:59Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#004GoldFramework
TencentCloudBase/CloudBase-MCP
What it is: CloudBase MCP is an MIT-licensed bridge that connects AI IDEs and coding agents to Tencent CloudBase so prompts can move beyond code generation into hosted app scaffolding, deployment, hosting, and backend wiring.
Why it matters: That matters because a lot of AI app demos still stop at generated code. A path from agent output to a live app with hosting and backend services is much more useful for builders who actually want to ship something.
Score8.9
Popularity86
RiskMedium
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum8.0
Maturity8.7
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
AI app buildersfull-stack developersTencent CloudBase users
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Agent Framework · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Medium risk from workflow or data-surface assumptions in 'AI App Deployment / Cloud'.
Caution flags
not hands-on testedrestricted permissionsdata handling risk
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-19T02:04:39Z · Updated: 2026-06-19T02:04:39Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#005GoldFramework
PrefectHQ/fastmcp
What it is: FastMCP is Prefect's Apache-2.0 Python framework for building MCP servers and clients with a high-level API, transport support, server composition, auth hooks, and deployment paths that reduce the boilerplate around exposing tools to agent clients.
Why it matters: Useful for developers who want to ship MCP integrations without hand-rolling transport plumbing, tool metadata, and server lifecycle code every time.
Score8.8
Popularity82
RiskConditional
VerificationTested in a bounded workflow
Score breakdown
Usefulness9.0
Novelty8.0
Momentum8.0
Maturity8.5
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
Python developers building MCP serversinternal platform teamsagent-tooling builders
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Mcp Infrastructure
AI relevance
Core AI
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: Tested in a bounded workflowA retained bounded workflow records setup, action, outcome, limitation, cleanup, and log evidence. Basis: Bounded representative workflow retained by RepoRadar verification harness. Last checked 2026-07-14.
First seen: 2026-06-19T11:09:11Z · Updated: 2026-06-19T11:09:11Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#006GoldTool
achiya-automation/safari-mcp
What it is: Safari MCP is an MIT-licensed macOS server that gives MCP agents 80 native Safari tools, keeps your real tabs and logins, and avoids a separate Chromium automation stack.
Why it matters: Useful for Mac users building browsing or QA workflows who want an agent to use the browser session they already trust instead of spinning up a separate headless environment.
Score8.7
Popularity63
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum8.0
Maturity8.1
Open-source/build8.4
Evidence7.2
Workflow potential9.8
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
Mac developersbrowser-agent buildersQA automation users
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Mcp Infrastructure
AI relevance
Core AI
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-19T16:06:11Z · Updated: 2026-06-19T16:06:11Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
What it is: Apache-2.0 licensed. The open-source search infrastructure for AI -- embedding database / search engine for LLM applications; dense + sparse + full-text + multi-vector search in one API; SQL-style metadata filtering, late-interaction / SPLADE / ColBERT rerankers, native LangChain / LlamaIndex / OpenAI / Claude / MCP integration; ships the official chroma-mcp server companion and Chroma Cloud / Enterprise
Why it matters: Most AI / ML engineers + RAG developers building production search-backed LLM applications have been either stitching together multiple vector DBs (Pinecone for dense + Elasticsearch for full-text + custom BM25 + custom reranker + custom metadata filtering -- high maintenance burden, no unified API), adopting a single-vendor vector DB (Pinecone, Weaviate, Qdrant Cloud) that locks-in the
Score8.7
Popularity0
RiskLow
VerificationTested in a bounded workflow
Score breakdown
Usefulness10.0
Novelty7.0
Momentum10.0
Maturity6.8
Open-source/build8.4
Evidence7.2
Workflow potential9.4
Setup ease8.8
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
AI developerstechnical foundersengineering leads
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Rag Knowledge · candidate; this does not change the authoritative category, score, verdict, or badges.
Useful-for basis
Broad catalog fallback while record-specific audience review is pending; not a record-specific recommendation.
Risk basis
No inherent risk flagged.
Caution flags
not hands-on tested
Verification: Tested in a bounded workflowA retained bounded workflow records setup, action, outcome, limitation, cleanup, and log evidence. Basis: Bounded representative workflow retained by RepoRadar verification harness. Last checked 2026-07-13.
First seen: 2026-07-08T19:09:29.100647+00:00 · Updated: 2026-07-08T19:09:29.100647+00:00 · Best-for basis: review pending fallback · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#008GoldMCP Server
CoplayDev/unity-mcp
What it is: CoplayDev/unity-mcp is an 11,232-star MIT bridge between AI assistants and the Unity Editor that exposes 47 focused MCP tools for scenes, assets, scripts, tests, profiling, and builds, with a fresh v10.0.0 release on 2026-06-30 and current support for Unity 2021.3 LTS through 6.x.
Why it matters: Useful for Unity teams that want AI-assisted scene building and editor automation from Claude, Codex, VS Code, local LLMs, or other MCP clients without inventing a custom bridge first.
Score8.7
Popularity1
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum9.0
Maturity6.8
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
Unity developers who want agent help beyond autocompleteTechnical artists and gameplay engineers prototyping scenes through MCPGame teams comparing editor-control bridges across Claude, Codex, Cursor, and local LLMs
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Coding Agent · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-07-01T21:08:53Z · Updated: 2026-07-01T21:08:53Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#009GoldTool
CursorTouch/Windows-MCP
What it is: Windows-MCP is an MIT-licensed MCP server that gives Claude, Codex, and other MCP clients practical Windows desktop control for file navigation, application control, UI automation, and QA-style computer-use workflows.
Why it matters: Useful for builders who want computer-use workflows on Windows without building their own desktop automation bridge from scratch.
Score8.7
Popularity74
RiskMedium
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty7.0
Momentum8.0
Maturity8.3
Open-source/build8.4
Evidence7.2
Workflow potential9.8
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
Windows power usersQA engineersAI agent builders
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Agent Framework · candidate; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Medium risk from workflow or data-surface assumptions in 'Browser / Computer Use'.
Caution flags
not hands-on testedrestricted permissionsdata handling risk
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-20T05:09:50Z · Updated: 2026-06-20T05:09:50Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#010GoldTool
czlonkowski/n8n-mcp
What it is: n8n-MCP is an MIT-licensed MCP server that gives AI assistants structured access to n8n node documentation, schemas, operations, AI-capable nodes, workflow templates, and verified community integrations.
Why it matters: Useful for builders who want assistants to assemble or explain n8n automations with real node-level context instead of hallucinating from stale docs or generic workflow examples.
Score8.7
Popularity78
RiskConditional
VerificationInstalled and tested
Score breakdown
Usefulness8.0
Novelty7.0
Momentum9.0
Maturity8.4
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
automation buildersn8n usersMCP users
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Coding Agent · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: Installed and testedA retained setup, action, outcome, limitation, cleanup, and log record exists. Basis: Hands-on setup or capability check retained by RepoRadar verification harness. Last checked 2026-07-13.
First seen: 2026-06-19T07:08:14Z · Updated: 2026-06-19T07:08:14Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#011GoldFramework
mcp-use/mcp-use
What it is: mcp-use is a full-stack TypeScript framework for building MCP apps and MCP servers for ChatGPT and Claude workflows.
Why it matters: It consolidates client/server plumbing, auth handling, and tool wiring so teams can stand up reliable MCP integrations without building protocol glue from scratch.
Score8.7
Popularity92
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum8.0
Maturity8.7
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
LLM platform teamsMCP buildersbackend engineers
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Mcp Infrastructure · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-18T01:11:20Z · Updated: 2026-06-18T01:11:20Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#012GoldTool
Bedrock AgentCore SDK: Apache-2.0 Framework-Agnostic Python SDK for Deploying AI Agents (Runtime, Memory, Auth, MCP Gateway, Code Interpreter, Browser, Observability)
What it is: Apache-2.0 framework-agnostic Python SDK for deploying AI agents to Amazon Bedrock AgentCore -- works with Strands / LangGraph / CrewAI / Autogen or any custom framework; agent primitives for runtime / memory / authentication / gateway (MCP) / code interpreter / browser / observability / identity with AWS-managed infrastructure and zero infrastructure management; ships the AG-UI protocol adapter
Why it matters: Most AI / ML engineers + agent developers deploying production AI agents today have been either (a) hand-rolling AWS infrastructure wiring for each agent primitive (Lambda + API Gateway + CloudWatch + IAM + Secrets Manager + DynamoDB for memory + a custom MCP gateway + custom sandboxed execution + custom browser automation), (b) adopting a single-vendor framework
Score8.6
Popularity0
RiskLow
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum9.0
Maturity6.7
Open-source/build8.4
Evidence7.2
Workflow potential9.3
Setup ease8.8
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
AI developerstechnical foundersengineering leads
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Agent Framework · exploratory; this does not change the authoritative category, score, verdict, or badges.
Useful-for basis
Broad catalog fallback while record-specific audience review is pending; not a record-specific recommendation.
Risk basis
No inherent risk flagged.
Caution flags
not hands-on tested
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-07-08T18:12:23.860014+00:00 · Updated: 2026-07-08T18:12:23.860014+00:00 · Best-for basis: review pending fallback · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#013GoldTool
ChromeDevTools/chrome-devtools-mcp
What it is: ChromeDevTools/chrome-devtools-mcp is an Apache-2.0 MCP server and CLI that lets coding agents inspect and control a live Chrome browser through DevTools. It gives agents screenshots, console logs, network traces, performance insights, and Puppeteer-backed browser actions instead of brittle blind UI automation.
Why it matters: Useful for teams using Claude, Cursor, Copilot, or other coding agents on web apps: wire it into a non-sensitive browser profile first, then use it to reproduce UI bugs, inspect console errors, and collect performance traces during agent-assisted debugging.
Score8.6
Popularity88
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum9.0
Maturity8.5
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
First seen: 2026-06-16T18:04:00Z · Updated: 2026-06-16T18:04:00Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#014GoldTool
Cline: Apache-2.0 Official Open-Source AI Coding Agent for VS Code + JetBrains + CLI (64,465*, MCP-Native, Plan/Act Dual Mode, Browser + Terminal + File Editing)
What it is: Apache-2.0 official open-source AI coding agent (the Cline fork lineage) for VS Code + JetBrains + CLI; 64,465* at verify time, last commit 2026-07-08; install via VS Code / JetBrains extension marketplace or `npm install -g cline`
Why it matters: Most software developers + AI engineers + coding-agent power users using AI coding assistants today have been either (a) locked into a single vendor (Cursor / GitHub Copilot / Windsurf / Cody / Codeium) with per-request markup + proprietary prompts + no model-provider flexibility + no MCP-native integration, (b) hand-rolling a custom agent loop with no canonical reference implementation + no
Score8.6
Popularity0
RiskMedium
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty7.0
Momentum10.0
Maturity6.7
Open-source/build8.4
Evidence7.2
Workflow potential9.3
Setup ease8.8
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
software engineersdeveloper-tool teams
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Coding Agent
AI relevance
Core AI
Risk basis
Medium risk from workflow or data-surface assumptions in 'Watchlist'.
Caution flags
not hands-on testedrestricted permissionsdata handling risk
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-07-09T03:20:29.724305+00:00 · Updated: 2026-07-09T03:20:29.724305+00:00 · Best-for basis: authoritative category default · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#015GoldMCP Server
docsagent/docsagent
What it is: docsagent/docsagent is a local-first documents intelligence engine with a native CLI and MCP server for indexing private folders, searching PDFs and Office files, and giving coding agents fast access to local document corpora without a cloud sync layer.
Why it matters: Useful for people who want agents to work against real local documents instead of copying files into chat or standing up a heavier enterprise knowledge stack first.
Score8.6
Popularity71
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum7.0
Maturity8.2
Open-source/build8.4
Evidence8.0
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
Developers and operators who want agents to search local document collectionsResearchers and analysts working with large private PDF or Office corporaTeams that need a local-first MCP server instead of a hosted knowledge base
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Mcp Infrastructure
AI relevance
Core AI
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-26T21:06:56Z · Updated: 2026-06-26T21:06:56Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#016GoldTool
Analytics MCP: Official Google Analytics MCP Server from the `googleanalytics` GitHub Org (13+ MCP Tools, Apache-2.0)
What it is: Apache-2.0 official Google Analytics MCP server -- Python package (`pip install analytics-mcp`, v0.6.0) that exposes 13+ MCP tools (`get_account_summaries`, `get_property_details`, `list_google_ads_links`, `run_report`, `run_funnel_report`, `get_custom_dimensions_and_metrics`, `run_realtime_report`, plus Property / User / Link admin tools) against the Google Analytics Admin API + Data API
Why it matters: Most marketers / growth analysts / product managers today who want an AI agent (Gemini CLI / Claude Code / Cursor / Windsurf) to query Google Analytics have been either (a) writing custom google-analytics-admin + google-analytics-data wrappers by hand (no canonical MCP surface), (b) reaching for non-MCP analytics libraries that require custom agent integration, or (c) asking the agent to
Score8.6
Popularity0
RiskLow
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum8.0
Maturity6.7
Open-source/build8.4
Evidence7.2
Workflow potential9.7
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
Marketers / growth analysts / product managers who want an AI agent (Gemini CLI / Claude Code / Cursor / Windsurf) to query Google Analytics via MCP + AI agent developers building analytics-awareMarketers + 13+-MCP-tool users that want the 13+ MCP tools (get_account_summaries, get_property_details, list_google_ads_links, run_report, run_funnel_report, get_custom_dimensions_and_metricsMarketers + pip-install-analytics-mcp-v0-6-0 users that want the `pip install analytics-mcp` v0.6.0 install -- the right pip-install-analytics-mcp primitive for any marketer who has been fighting
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Data Tooling · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
No inherent risk flagged.
Caution flags
not hands-on tested
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-07-08T13:35:39+00:00 · Updated: 2026-07-08T13:35:39+00:00 · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#017GoldTool
HelpCode-ai/anythingmcp
What it is: AnythingMCP is an AGPL-licensed self-hosted gateway that turns REST, SOAP/WSDL, GraphQL, SQL/NoSQL, and existing MCP servers into connectors for Claude, ChatGPT, Gemini, Copilot, and Cursor, with OAuth2, RBAC, audit logs, and a growing library of prebuilt adapters.
Why it matters: Useful for teams that want to expose internal systems to AI clients without hand-writing a fresh connector for every API or database.
Score8.6
Popularity74
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum7.0
Maturity8.2
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease4.2
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
First seen: 2026-06-19T13:07:16Z · Updated: 2026-06-19T13:07:16Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#018GoldMCP Server
microsoft/playwright-mcp
What it is: microsoft/playwright-mcp is an official Apache-2.0 Playwright MCP server that gives agents deterministic browser control through accessibility snapshots instead of screenshot guessing, with broad install docs across Codex, Claude Code, VS Code, Cursor, Goose, Docker, and standalone HTTP deployment paths.
Why it matters: Useful for builders who need browser automation that is structured, widely documented, and grounded in an official Playwright-maintained MCP surface rather than a fragile wrapper.
Score8.6
Popularity34385
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum8.0
Maturity9.2
Open-source/build8.4
Evidence8.0
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
Agent builders who need reliable browser control without a vision-only stackDevelopers wiring web automation into Codex, Claude Code, VS Code, Cursor, or similar MCP clientsTeams standardizing on an official browser MCP instead of a community wrapper
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Coding Agent · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-26T15:03:24Z · Updated: 2026-06-26T15:03:24Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#019GoldTool
smart-mcp-proxy/mcpproxy-go
What it is: MCPProxy is an MIT-licensed single-binary proxy for MCP servers that federates large tool surfaces behind retrieval, quarantines newly discovered servers, and can run external security scanners before approval. It is built for teams hitting tool-count limits or worrying about unsafe MCP sprawl.
Why it matters: Useful for MCP-heavy agent setups that need a policy checkpoint instead of blindly attaching more servers: start with a small internal tool set, verify the quarantine and approval flow, then decide whether it really reduces token and safety overhead.
Score8.6
Popularity52
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum7.0
Maturity7.8
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease4.2
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
MCP power usersagent platform teamssecurity-minded developers
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Security · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-19T19:05:06Z · Updated: 2026-06-19T19:05:06Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#020GoldTool
SylphxAI/pdf-reader-mcp
What it is: PDF Reader MCP is an MIT-licensed local-first PDF server for AI agents with search, OCR, page-region crops, accessibility checks, and structured extraction for real document workflows.
Why it matters: Useful for teams building document-heavy AI workflows who need something more production-ready than ad hoc PDF parsing scripts.
Score8.6
Popularity72
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty7.0
Momentum8.0
Maturity8.2
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
document workflow buildersRAG developersresearch teams
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Data Tooling · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-19T16:06:11Z · Updated: 2026-06-19T16:06:11Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#021GoldTool
Desktop Commander MCP: MIT MCP Server for Claude Desktop with Terminal Control + File System Search + Diff File Editing (6,377*, npm `npx @wonderwhy-er/desktop-commander`, Audit Logged)
What it is: MIT MCP server for Claude Desktop with terminal control + file system search + read/write/edit files (text/Excel/PDF/DOCX) + recursive directory listing + vscode-ripgrep-based code/text search + diff file editing + comprehensive audit logging; 6,377* at verify time, last commit 2026-06-26; install `npx @wonderwhy-er/desktop-commander@latest setup`
Why it matters: Most AI engineers + Claude Desktop users + MCP-client developers needing terminal / file-system access today have been either (a) using Claude Desktop's built-in file tools (limited to specific folders, no recursive code search, no terminal access by default), (b) adopting a single-vendor MCP server (Smithery / Anthropic official servers / third-party forks) that locks-in the deployment model and
Score8.6
Popularity0
RiskMedium
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty7.0
Momentum10.0
Maturity6.7
Open-source/build8.4
Evidence7.2
Workflow potential9.3
Setup ease8.8
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
AI developerstechnical foundersengineering leads
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Agent Framework · exploratory; this does not change the authoritative category, score, verdict, or badges.
Useful-for basis
Broad catalog fallback while record-specific audience review is pending; not a record-specific recommendation.
Risk basis
Medium risk from workflow or data-surface assumptions in 'Watchlist'.
Caution flags
not hands-on testedrestricted permissionsdata handling risk
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-07-09T01:23:15.280646+00:00 · Updated: 2026-07-09T01:23:15.280646+00:00 · Best-for basis: review pending fallback · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#022GoldSecurity Tool
abluva-research/mcp-trust-plane
What it is: MCP Trust Plane is an Apache-2.0 filter framework for Model Context Protocol traffic that lets teams allow, block, redact, truncate, or modify tool calls and responses through pluggable HTTP filters.
Why it matters: Useful for teams that want to add policy and safety controls around MCP-connected agents without rewriting the upstream MCP server or the agent runtime itself.
Score8.5
Popularity62
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness8.0
Novelty8.0
Momentum7.0
Maturity7.9
Open-source/build8.4
Evidence7.2
Workflow potential9.3
Setup ease4.2
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
teams deploying MCP serversagent platform engineerssecurity-conscious automation builders
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Security
AI relevance
Core AI
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-19T08:06:53Z · Updated: 2026-06-19T08:06:53Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#023GoldMCP Server
agent-sh/agent-workspace-linux
What it is: agent-sh/agent-workspace-linux is an MIT Linux MCP runtime that gives coding agents a hidden X11 desktop, a workspace-owned browser, a live viewer, and optional daemon-enforced network, mount, and app ceilings, so GUI and browser tasks can run in a separate agent workspace instead of hijacking the user's real desktop session.
Why it matters: Useful for developers and ops teams who need real browser or desktop automation with a clearer safety boundary than pointing an agent at their live machine.
Score8.5
Popularity88
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum7.0
Maturity8.5
Open-source/build8.4
Evidence8.0
Workflow potential10.0
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
Teams testing browser QA or GUI automation without surrendering the host desktopDevelopers building MCP-based computer-use flows with a clearer boundary modelOps and support workflows that need a throwaway browser profile inside an agent-owned workspace
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Mcp Infrastructure
AI relevance
Core AI
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-26T17:08:14Z · Updated: 2026-06-26T17:08:14Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#024GoldTool / MCP
firecrawl/firecrawl-mcp-server
What it is: Firecrawl's official MIT-licensed MCP server gives Claude, Cursor, and other clients a clean way to search, scrape, and extract web content through Firecrawl without custom glue code in every workflow.
Why it matters: Useful for people building agents that need dependable website retrieval and extraction instead of brittle one-off scraping scripts.
Score8.5
Popularity58
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty6.0
Momentum8.0
Maturity7.8
Open-source/build8.4
Evidence7.2
Workflow potential9.6
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
agent buildersresearch automationsupport teams extracting docs
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Data Tooling · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-20T06:02:16Z · Updated: 2026-06-20T06:02:16Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#025GoldTool / Platform
IBM/mcp-context-forge
What it is: ContextForge is IBM's open-source, Apache-2.0, 3,914-star registry and proxy that federates MCP, A2A, and REST/gRPC APIs behind one clean endpoint — ships Tools Gateway (MCP, REST, gRPC-to-MCP translation, TOON compression), Agent Gateway (A2A, OpenAI-/Anthropic-compatible routing), API Gateway (rate limiting, auth, retries, reverse proxy), 40+ plugins, and OpenTelemetry tracing. PyPI + Docker
Why it matters: Useful for platform and infra teams that need a vendor-neutral, production-grade AI gateway to federate the explosion of MCP servers, A2A agents, and REST/gRPC APIs behind one observable, governable endpoint — without standing up a custom Envoy/Istio stack or wiring their own OAuth/registry.
Score8.5
Popularity86
RiskNone
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum8.0
Maturity8.4
Open-source/build8.4
Evidence7.2
Workflow potential9.6
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
platform teams that need to federate the explosion of MCP servers, A2A agents, and REST/gRPC APIs behind one endpointinfra teams that need a vendor-neutral, observable, governable AI gateway without writing their own Envoy/Istio stacksecurity teams that need OAuth 2.0 (client + server modes), rate limiting, retries, and reverse proxy for AI traffic
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Inference Runtime · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
No inherent risk flagged.
Caution flags
not hands-on tested
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-17T16:00:00Z · Updated: 2026-06-17T16:00:00Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
What it is: Apache-2.0 all-in-one open-source backend platform for agentic coding with 8 surface primitives (Authentication / Database / Storage / Edge Functions / Model Gateway / Compute / Deployment); MCP Server (self-hosted and cloud) exposes InsForge operations as tools any MCP-compatible agent can call; CLI + Skills (cloud only) interface lets agents invoke operations directly from the terminal; reading
Why it matters: Most coding-agent developers today who want a backend for their Claude Code / Codex / Cursor / open-source coding-agent integrations have been either (a) wiring up Supabase + Clerk + Stripe + Vercel + AWS S3 + their own auth provider -- which is a 3-week setup with 6 vendor lock-ins, or (b) hand-rolling Postgres + NextAuth + a deployment pipeline + storage -- which is a 3-month build.
Score8.5
Popularity0
RiskLow
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum9.0
Maturity6.6
Open-source/build8.4
Evidence7.2
Workflow potential9.6
Setup ease8.8
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
Coding agent developers + full-stack developers + AI engineers building Claude Code / Codex / Cursor / open-source coding-agent integrations + indie developers + small teams + any developer wantingCoding agent developers + MCP-tool-call users that want the MCP Server (self-hosted and cloud) exposing InsForge operations as tools any MCP-compatible agent can call -- the right MCP-tool-callCoding agent developers + CLI / Skills terminal-invocation users that want the CLI + Skills interface for terminal-direct invocation -- the right terminal-invocation primitive for any coding-agent
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Coding Agent · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
No inherent risk flagged.
Caution flags
not hands-on tested
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-07-08T10:08:51+00:00 · Updated: 2026-07-08T10:08:51+00:00 · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#027GoldMCP Traffic Inspector
kerlenton/mcpsnoop
What it is: mcpsnoop is a 143-star, 5-subscriber MIT Go CLI that sits in the real data path between an MCP client (Claude Desktop, Cursor, Claude Code) and an MCP server and shows every JSON-RPC frame in a live terminal UI — 'Wireshark for MCP' that wraps the server command and exposes the actual tool calls the real client made
Why it matters: Useful for MCP server authors, MCP client authors, agent harness developers, and anyone debugging 'the tool was supposed to be called but never was' or 'the call hung' mysteries in Claude Desktop / Cursor / Claude Code — the canonical replacement for `tail -f /tmp/some.log` and the right primitive for closing the gap between an MCP server's intended API and the calls a real client actually makes.
Score8.5
Popularity1
RiskLow
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty9.0
Momentum7.0
Maturity6.6
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease8.8
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
MCP server authors debugging 'the tool silently wasn't called / capabilities don't line up / the call hangs' mysteries in production — replace the server command in your Claude Desktop / Cursor /MCP client authors reverse-engineering what a real client actually sends vs. what the docs say — run `mcpsnoop -- <server-command>` against your own server and let the TUI show what the client'sAgent harness developers building a Claude Desktop / Cursor / Claude Code workflow that depends on an MCP server — point the harness at `mcpsnoop --` wrapped server, watch the live traffic while
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Mcp Infrastructure
AI relevance
Core AI
Risk basis
No inherent risk flagged.
Caution flags
not hands-on tested
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-07-03T22:30:00Z · Updated: 2026-07-03T22:30:00Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#028GoldTool
shuymn/gh-mcp
What it is: gh-mcp is a focused MIT-licensed GitHub CLI extension that launches the GitHub MCP server with your existing gh authentication, cutting out the manual token setup that often keeps GitHub-connected agent workflows theoretical.
Why it matters: That matters because GitHub MCP is much more useful when a developer can stand it up quickly with the credentials they already use, rather than hunting down and managing separate personal access tokens for every test.
Score8.5
Popularity66
RiskConditional
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty7.0
Momentum7.0
Maturity8.0
Open-source/build8.4
Evidence7.2
Workflow potential10.0
Setup ease8.8
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
developers using GitHub MCPCLI-first GitHub usersagent workflow builders
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Agent Framework · exploratory; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Some conditional risk identified; confirm assumptions and environment before adopting in production.
Caution flags
not hands-on testedsensitivity
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-19T02:04:39Z · Updated: 2026-06-19T02:04:39Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#029GoldTool
taylorwilsdon/google_workspace_mcp
What it is: Google Workspace MCP Server packages Gmail, Calendar, Drive, Docs, Sheets, Slides, Chat, Forms, Tasks, and related tools behind one MCP server and CLI so agents can work across a real workspace instead of isolated apps.
Why it matters: This is useful because assistants become more practical when they can connect mail, calendar, files, and documents in one flow. It also concentrates real account permissions, so scope control matters.
Score8.5
Popularity84
RiskMedium
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty8.0
Momentum8.0
Maturity8.4
Open-source/build8.4
Evidence7.2
Workflow potential9.3
Setup ease6.4
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
Google Workspace power usersoperations teamsautomation builders
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Other
AI relevance
Core AI
Research topic (model-assisted)
Business Workflow · candidate; this does not change the authoritative category, score, verdict, or badges.
Risk basis
Medium risk from workflow or data-surface assumptions in 'AI Automation'.
Caution flags
not hands-on testedrestricted permissionsdata handling risk
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-18T23:06:24Z · Updated: 2026-06-18T23:06:24Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.
G#030GoldLocal MCP gateway that
tsouth89/conduit
What it is: tsouth89/conduit is the MIT local MCP gateway that sits between every AI client (Claude Desktop / Claude Code, Cursor, VS Code / GitHub Copilot, Codex CLI / Codex app) and the user's installed MCP servers, and lazy-collapses all of them into 3 meta-tools (`search_tools`, `describe_tool`, `call_tool`) the agent discovers on demand — for ~90% fewer tokens per turn versus eagerly registering every
Why it matters: Useful for AI coding-agent power users (Claude Desktop / Claude Code, Cursor, VS Code / GitHub Copilot, Codex CLI / Codex app) who run multiple MCP servers and hit the context-bloat wall — eager registration of all N×M tools blows past the context window in seconds, and the agent wastes tokens re-reading tool descriptions it never uses; conduit sits between the client and every MCP server and
Score8.5
Popularity35
RiskNone
VerificationDiscovered
Score breakdown
Usefulness9.0
Novelty9.0
Momentum7.0
Maturity7.4
Open-source/build8.4
Evidence8.0
Workflow potential10.0
Setup ease8.8
Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.
Useful for
AI coding-agent power users (Claude Desktop / Claude Code, Cursor, VS Code / GitHub Copilot, Codex CLI / Codex app) who run multiple MCP servers and hit the context-bloat wall — eager registrationUsers who care about **credential security** — auth profiles are first-class, written to the **OS keychain** (macOS Keychain / Windows Credential Manager / Linux Secret Service via D-Bus) ratherUsers who need **per-server ACLs** — allow/deny rules are ACL-style regexes against tool names (e.g. `allow = ["^search", "^read_"]`, `deny = ["^delete_", "^write_secrets"]`) so a coding agent can
Skill fit
Power-user friendly
Setup
Moderate
Code
Optional
Authoritative category
Mcp Infrastructure
AI relevance
Core AI
Risk basis
No inherent risk flagged.
Caution flags
not hands-on tested
Verification: DiscoveredSource captured; no editorial or hands-on claim. Basis: Automated discovery and source capture. Last checked 2026-08-21.
First seen: 2026-06-25T14:30:00Z · Updated: 2026-06-25T14:30:00Z · Best-for basis: canonical useful for · AI relevance: Direct evidence of LLMs, agents, models, RAG, evals, model serving, or AI-native workflows.