Use this MCP server list to find a connection for a specific task. If you need repository context, start with GitHub or GitLab. For public documentation, look at Microsoft Learn or Context7. A database question needs a server that can reach the relevant database with the permissions you intend to grant.
The 60 entries below are grouped into eight categories. Each includes a suggested task, connection type, authentication and a limitation worth checking before setup. Download the comparison CSV if you want to keep a shortlist outside the browser.
Documentation reviewed on September 27, 2026. The examples are proposed tasks: WexAI has not authenticated to these 60 servers, inspected their live tool lists or executed the examples. “Documented” means the provider describes the server; it is not a general-availability or reliability guarantee. Beta, developer preview and reference implementations have separate labels. Your account, client and region can change what is available.
What is an MCP server?
Model Context Protocol (MCP) is a standard for connecting AI applications to external context and tools. An MCP server exposes capabilities to a client within an AI application. Those capabilities may read a document, query a database or take an action in another service. The protocol overview explains the relationship between hosts, clients and servers.
A model name alone does not establish compatibility. Check the application running the model: its supported transport, authentication flow and permission controls must match the server. A local server usually runs as a process on your computer; a remote server runs behind a network endpoint. Local execution can still call a paid external API or access sensitive files.
How to choose from this list
Write down the operation before choosing a product. “Read the schema of a test database” gives you a useful access boundary. “Connect all our tools” does not specify what the assistant is allowed to do.
- Identify the task and the data it requires. Select the category that owns those inputs.
- Check which tools perform the operation. Reading an issue and creating a pull request need different permissions.
- Choose the connection and authentication your client supports. Use the provider’s current setup instructions rather than copying a configuration from a different client.
- Restrict the account, project, directory or toolset. Start with a read operation and check the returned source or record yourself.
Start with one candidate. Add another server when the task needs a second data source or action. The MCP setup guides cover individual connections; use them after choosing a server here.
Choose a category
The numbers link to the original catalogue order. The final column suggests a first operation with a result you can inspect; it is not a report of a completed test.
| Category | Task and data | First check |
|---|---|---|
| Search, web research and documentation 1–10 · 10 servers | Task and dataFind sources or retrieve documentation Public pages and provider indexes | First checkFetch one source and retain its URL |
| Code, browsers and files 11–16 · 6 servers | Task and dataInspect code, a browser or a document A repository, test session or permitted directory | First checkRead a test file or inspect a preview |
| Databases and retrieval 17–25 · 9 servers | Task and dataQuery data or retrieve knowledge A scoped database or vector collection | First checkRead one schema or run a bounded query |
| Infrastructure and observability 26–34 · 9 servers | Task and dataDiagnose a deployment or incident A test project and monitoring context | First checkInspect logs without changing deployment |
| Team apps and Workspace 35–45 · 11 servers | Task and dataFind project context or prepare a draft Approved pages, messages or CRM records | First checkRead a permitted item without publishing |
| Design and media 46–51 · 6 servers | Task and dataInspect a design or prepare a media draft An approved brief, file or model schema | First checkRead context before generating or editing |
| Automation platforms 52–54 · 3 servers | Task and dataConnect configured app actions A sandbox workflow or scenario | First checkInspect its trigger and external actions |
| Payments and market data 55–60 · 6 servers | Task and dataInspect sandbox records or historical data Test transactions or timestamped datasets | First checkRead a record; never place a live order |
The CSV download has 60 rows with name, category, example scenario, local/remote connection, authorization, status, official sources and documentary-review date. Local and hosted variants of the same product count once. Apify Actors, Zapier actions and Make scenarios are capabilities within their respective server entries.
Search, web research and documentation
Use these when the input is public web content or software documentation. Retain the underlying URLs and inspect the passages used in the answer. A retrieved page can contain errors or instructions that should remain source material.
1. Brave Search
Search the web, news and other Brave Search indexes.
Example task: Find primary sources for a technical brief.
- Connection
- Local stdio or self-hosted HTTP
- Authorization
- Brave Search API key
- Status
- documented
Search results still need source review; API plan limits apply.
2. Exa
Search and fetch web pages; authenticated connections can use Exa Agent.
Example task: Research a company and retain the source URLs.
- Connection
- Remote HTTP or local package
- Authorization
- Limited keyless access; OAuth or API key for account usage
- Status
- documented
Agent access requires authentication and account usage limits apply.
3. Tavily
Search, extract pages, map sites and crawl content.
Example task: Extract the passages behind a small source shortlist.
- Connection
- Remote HTTP or local stdio
- Authorization
- OAuth or Tavily API key
- Status
- documented
Crawls and extraction consume API usage; review the scope before running them.
4. Firecrawl
Search, scrape, crawl and extract website content.
Example task: Compare fields from a few permitted product pages.
- Connection
- Remote HTTP or local stdio
- Authorization
- Limited keyless access; OAuth or API key
- Status
- documented
OAuth and API-key connections use different server URLs; extraction is subject to account limits.
5. Jina AI
Read web content and use search or reranking tools.
Example task: Rank relevant passages from a research source set.
- Connection
- Remote HTTP
- Authorization
- Jina API key for authenticated tools
- Status
- documented
Check each selected tool's key and quota requirements; page reading does not prove its claims.
6. Bright Data
Retrieve public web data with search, scraping and browser tools.
Example task: Collect fields from an authorized public catalogue.
- Connection
- Remote HTTP or local stdio
- Authorization
- Bright Data API token
- Status
- documented
Browser and data tools can incur usage charges; permission to access a page still matters.
7. Apify Actors
Discover Actors, run selected Actors and retrieve their datasets.
Example task: Find an extraction Actor and inspect its output schema before a run.
- Connection
- Remote HTTP or local stdio
- Authorization
- OAuth or Apify token; limited discovery without authentication
- Status
- documented
Actors can cost money and execute work; each Actor is not a separate server in this list.
8. Perplexity API
Use Perplexity search and research tools through its API.
Example task: Prepare a sourced overview for later fact checking.
- Connection
- Remote HTTP or local stdio
- Authorization
- OAuth or Perplexity API key
- Status
- documented
API usage has its own billing; a consumer subscription is not proof of API entitlement.
9. Microsoft Learn
Search Microsoft documentation and retrieve documentation or code examples.
Example task: Find the documented prerequisites for an Azure feature.
- Connection
- Remote HTTP
- Authorization
- No account authentication for public documentation
- Status
- documented
This is a documentation server, not access to your Azure resources.
10. Context7
Retrieve documentation and examples for software libraries.
Example task: Check a coding answer against the intended library version.
- Connection
- Remote HTTP or local stdio
- Authorization
- Context7 API key recommended; the setup helper can obtain a key through OAuth
- Status
- documented
Version-specific documentation can help review a draft; it does not establish that generated code works.
Code, browsers and files
Choose the scope before attaching a repository, browser session or directory. A disposable project and a test browser account make it easier to review proposed changes without risking personal files or live services.
11. GitHub
Read repository context and work with issues and pull requests.
Example task: Inspect an issue and draft a patch for review.
- Connection
- Remote HTTP or local stdio
- Authorization
- OAuth or GitHub personal access token
- Status
- documented
Limit repositories and toolsets; read-only mode excludes write tools.
12. GitLab
Expose GitLab project context and supported issue, merge-request and CI tools.
Example task: Summarize the context around a merge request.
- Connection
- Remote instance HTTP endpoint
- Authorization
- OAuth
- Status
- beta
Beta availability depends on the GitLab version and instance configuration.
13. Playwright
Control a browser through navigation, page snapshots and interaction tools.
Example task: Check a demo onboarding flow in a test account.
- Connection
- Local stdio or self-hosted server
- Authorization
- No service API key; browser sessions carry website access
- Status
- documented
Clicks and form submissions can change real accounts; use an isolated test session.
14. Chrome DevTools
Inspect browser console, network activity and performance.
Example task: Investigate a failed request on a local preview.
- Connection
- Local Chrome connection
- Authorization
- No separate service key; access to the browser debugging session
- Status
- documented
A connected browser may expose session data; avoid attaching a personal authenticated session.
15. Filesystem
Read, search and modify files in permitted directories.
Example task: Inspect a disposable project folder before proposing edits.
- Connection
- Local stdio
- Authorization
- Operating-system permissions and configured allowed directories
- Status
- reference
Reference implementation; some tools overwrite or move files. Roots can change the permitted directories.
16. MarkItDown
Convert supported files and URIs into Markdown.
Example task: Convert an approved document into text for review.
- Connection
- Local Python server
- Authorization
- No service account for basic local conversion
- Status
- documented
Conversion can lose layout. The local server inherits your OS privileges and has no authentication; keep its HTTP mode on loopback.
Databases and retrieval
Use a test database, limited user and bounded query for the first connection. Inspect which tools can write as well as which tools can read. A client-side instruction to avoid writes is not a substitute for database permissions.
17. Supabase
Inspect database schemas and use SQL, migration and development tools.
Example task: Review a staging schema with read-only queries.
- Connection
- Remote HTTP or local development setup
- Authorization
- OAuth; personal access token for supported automation
- Status
- documented
Use project_ref and read_only=true; developer credentials are not end-user app permissions.
18. Neon
Work with Postgres projects, branches, schemas and queries.
Example task: Inspect a test branch before a migration proposal.
- Connection
- Remote HTTP or local stdio
- Authorization
- OAuth or Neon API key
- Status
- documented
Choose read-only access and one project; write grants can create resources or run migrations.
19. MongoDB
Inspect collections and run supported MongoDB queries or Atlas operations.
Example task: Review an aggregation against a test database.
- Connection
- Local server; managed Atlas option where supported
- Authorization
- Database connection credentials or supported Atlas authorization
- Status
- documented
Configure read-only access explicitly; do not assume every deployment starts read only.
20. Redis
Read and manage Redis data types and search operations.
Example task: Inspect a sandbox cache for unexpected keys.
- Connection
- Local stdio or self-hosted server
- Authorization
- Redis credentials and ACLs
- Status
- documented
Use a restricted Redis user; write tools can change or remove data.
21. Elastic Agent Builder
Expose configured Elastic tools, including search and ES|QL.
Example task: Group relevant errors from an approved log index.
- Connection
- Remote deployment endpoint
- Authorization
- API key or supported deployment OAuth
- Status
- documented
Tool and index access depend on the deployment's permissions and enabled configuration.
22. ClickHouse
Inspect schemas and run ClickHouse analytical queries.
Example task: Compare event totals in a test analytics database.
- Connection
- Local stdio or self-hosted HTTP
- Authorization
- ClickHouse database credentials; HTTP also needs configured server authentication
- Status
- documented
Keep the database user read only unless writes are deliberately required.
23. MotherDuck
Query analytical data and inspect schemas through a hosted server.
Example task: Prepare a data summary with the SQL used to obtain it.
- Connection
- Remote HTTP
- Authorization
- MotherDuck OAuth
- Status
- documented
The query_rw tool can change data and schemas; exclude it from a read-only task.
Official documentation · Official source 2 · Official source 3
24. Qdrant
Search and store information in Qdrant collections.
Example task: Retrieve passages from a small test knowledge base.
- Connection
- Local stdio or self-hosted server
- Authorization
- Qdrant URL and API key where required; local storage option
- Status
- documented
Storage tools write to collections; enable the documented read-only setting when appropriate.
25. Pinecone
Manage indexes and use vector search, data operations and reranking.
Example task: Compare retrieval results in a disposable index.
- Connection
- Local stdio
- Authorization
- Pinecone API key
- Status
- documented
Index management and upserts can change data; account permissions and quotas still apply.
Infrastructure and observability
Begin with diagnostic context from a test project. Logs may contain secrets or customer data. Deployment and infrastructure changes need an explicit review point beyond the initial connection.
26. Cloudflare API
Search the Cloudflare API and execute selected API operations through Code Mode.
Example task: Inspect a configuration and draft a change plan.
- Connection
- Remote HTTP
- Authorization
- Cloudflare OAuth or scoped API token
- Status
- documented
Execution can modify infrastructure; choose minimal token permissions and approve changes separately.
27. Vercel
Inspect Vercel projects, deployments and diagnostic context.
Example task: Investigate a failed preview deployment.
- Connection
- Remote HTTP
- Authorization
- Vercel OAuth
- Status
- beta
Beta; account and team access determine available operations. Do not deploy production by accident.
28. Netlify
Connect an agent to Netlify site and deployment operations.
Example task: Inspect a test site's deployment context.
- Connection
- Remote HTTP or local setup
- Authorization
- Netlify account authorization or token for the selected setup
- Status
- documented
Deployment tools can publish changes; keep the first task on a preview site.
29. Render
Inspect services, logs and metrics, with supported service and deployment actions.
Example task: Find the cause of a test service's startup failure.
- Connection
- Remote HTTP or local server
- Authorization
- Render OAuth or API key for the selected setup
- Status
- documented
Service changes and redeployments have external effects; supported tools do not cover every admin action.
30. Terraform
Look up Terraform providers and modules, with optional HCP Terraform tools.
Example task: Draft a provider configuration from its registry documentation.
- Connection
- Local server or self-hosted HTTP
- Authorization
- Public registry lookup needs no HCP token; private tools require credentials
- Status
- documented
Registry lookup alone does not run terraform apply; review any enabled HCP toolsets separately.
31. Sentry
Search errors, inspect performance and triage issues.
Example task: Compare an error with its release context.
- Connection
- Remote HTTP
- Authorization
- Sentry OAuth
- Status
- documented
Scope the server to one organization or project; an issue summary is not a validated fix.
32. Grafana
Use configured Grafana dashboard, metric and log tools.
Example task: Compare logs and metrics around an incident.
- Connection
- Local stdio or self-hosted server
- Authorization
- Grafana service account token and configured data-source access
- Status
- documented
Dashboard or other write tools depend on permissions; use a limited service account.
33. Datadog
Retrieve monitoring context such as logs, metrics and traces.
Example task: Collect diagnostic context for an incident timeline.
- Connection
- Remote regional endpoint
- Authorization
- OAuth; personal/service access token or API and application keys
- Status
- documented
Choose the endpoint for your Datadog site and restrict access to the required monitoring resources.
Official documentation · Official source 2 · Official source 3 · Official source 4
34. Docker Hub
Work with Docker Hub repository and image information.
Example task: Compare image tags before choosing a dependency.
- Connection
- Local Docker MCP setup
- Authorization
- Docker Hub username and personal access token for account operations
- Status
- documented
Docker Hub registry tools are not permission to manage the local Docker Engine.
Team apps and Workspace
These servers bring project and communication context into an assistant. Authorize only the workspace content needed for the task. Drafting a reply or status update should remain separate from sending or publishing it.
Google Drive, Gmail and Calendar are three distinct Google Workspace servers. Their developer preview requires programme membership and Cloud/OAuth configuration, according to Google’s configuration guide. Existing consumer connectors do not establish access to these developer endpoints.
35. Notion
Search, read, create and update accessible Notion content.
Example task: Draft a project decision note from approved pages.
- Connection
- Remote HTTP
- Authorization
- Notion OAuth
- Status
- documented
The server can update pages; workspace and user permissions define the accessible content.
36. Slack
Search Slack context and use supported message operations.
Example task: Summarize an approved channel without publishing a message.
- Connection
- Remote Streamable HTTP
- Authorization
- Slack OAuth through a supported client or registered app
- Status
- documented
Custom clients need Slack's app setup; do not assume dynamic registration or legacy SSE support.
37. Atlassian
Use Jira, Confluence and other supported Atlassian context and actions.
Example task: Relate a requirement page to its project issues.
- Connection
- Remote HTTP
- Authorization
- OAuth or supported API-token authentication
- Status
- documented
User permissions apply; the combined Atlassian server counts once, not once per product.
38. Linear
Read and manage Linear issues and projects.
Example task: Turn an approved plan into a draft issue list.
- Connection
- Remote HTTP
- Authorization
- OAuth or supported bearer token/API key
- Status
- documented
The default endpoint allows writes; use /mcp/readonly or a read-only OAuth grant for inspection.
39. Asana v2
Search and manage supported tasks and project context.
Example task: Prepare a project status summary for review.
- Connection
- Remote Streamable HTTP
- Authorization
- Asana OAuth
- Status
- documented
Use the v2 endpoint; organization app controls and client support can affect access.
40. monday.com Platform
Read and manage monday.com workspace data through Platform tools.
Example task: Summarize a board and propose status changes.
- Connection
- Remote Streamable HTTP
- Authorization
- OAuth or personal API token
- Status
- documented
Platform MCP differs from Apps MCP; legacy SSE is unsupported and API limits apply.
41. HubSpot Remote
Retrieve permitted CRM context, including contacts, companies and deals.
Example task: Prepare a deal summary using permitted CRM records.
- Connection
- Remote server
- Authorization
- HubSpot MCP connector OAuth with PKCE; custom integrations need an OAuth client
- Status
- documented
Sensitive data and some objects have access restrictions; the developer CLI server is a different product.
42. Intercom
Find and read conversations, contacts and companies, and use help-center article tools.
Example task: Identify questions that may justify a help-center update.
- Connection
- Remote Streamable HTTP
- Authorization
- Intercom OAuth or bearer access token
- Status
- documented
The documentation limits support to US-hosted workspaces; help-center article tools can write.
43. Google Drive
Search and read files, with documented file and metadata operations.
Example task: Compare permitted project source documents.
- Connection
- Remote Google Workspace endpoint
- Authorization
- Google OAuth, Cloud project, enabled APIs and MCP services
- Status
- developer preview
Developer Preview Programme membership is required; this is not universal Drive connector availability.
Official documentation · Official source 2 · Official source 3
44. Gmail
Search and read messages or threads and create drafts.
Example task: Draft a reply to a selected test conversation.
- Connection
- Remote Google Workspace endpoint
- Authorization
- Google OAuth, Cloud project, enabled APIs and MCP services
- Status
- developer preview
Developer preview; the reviewed tools reference lists draft creation, not a send tool.
Official documentation · Official source 2 · Official source 3
45. Google Calendar
Read calendar context and check availability; the reference also lists event mutations.
Example task: Find a free slot without creating an event.
- Connection
- Remote Google Workspace endpoint
- Authorization
- Google OAuth, Cloud project, enabled APIs and MCP services
- Status
- developer preview
Developer preview; setup scopes are read/free-busy while the reference includes writes. Event creation is not verified here.
Official documentation · Official source 2 · Official source 3
Design and media
Inspect the source design, brief or model input before generation. Distinguish reading design context from changing a canvas, running a paid model or publishing a site.
46. Figma
Retrieve design context and use supported design tools.
Example task: Inspect a screen's components before implementing it.
- Connection
- Remote server or desktop server
- Authorization
- Figma OAuth/account access for the chosen connection
- Status
- documented
Tools, client support and usage limits vary; design access does not guarantee an accurate implementation.
47. Canva
Use Canva's design connector to find, create or edit supported designs.
Example task: Prepare a draft design from an approved brief.
- Connection
- Remote HTTP
- Authorization
- Canva OAuth
- Status
- documented
This entry is the design connector, not the separate developer MCP; plan and tool limits apply.
48. Webflow
Work with site structure, styles, components, assets and CMS content.
Example task: Prepare a draft CMS item on a test site.
- Connection
- Remote server
- Authorization
- Webflow OAuth
- Status
- documented
Some visual or selection operations require the Designer and Bridge App; publishing needs separate approval.
49. fal Run
Search models and use supported media-generation APIs.
Example task: Inspect a model's inputs before generating one approved draft.
- Connection
- Remote HTTP
- Authorization
- OAuth or fal API key
- Status
- documented
Generation can incur model charges; this is Run MCP, not the separate platform-management server.
50. ElevenLabs Hosted
Create, configure and manage agents in an ElevenLabs workspace.
Example task: Inspect a test voice agent's configuration.
- Connection
- Remote regional HTTP endpoint
- Authorization
- ElevenLabs OAuth for the workspace's region
- Status
- documented
Use the endpoint for your workspace's region. Hosted tools are for agent management; do not inherit features from the archived local repository.
51. Hugging Face
Search Hub models, datasets, Spaces, papers and documentation; selected tools can run compute.
Example task: Compare model documentation for a small prototype.
- Connection
- Remote Hub server
- Authorization
- Hugging Face account and client-specific settings
- Status
- documented
Space tools and Jobs can execute code or use compute; review each selected tool before enabling it.
Automation platforms
Review the downstream actions of a workflow before exposing it as a tool. A short MCP call can start a longer process that sends messages, updates records or consumes usage credits.
52. Zapier
Expose configured actions across connected applications.
Example task: Prepare an approved handoff between two test apps.
- Connection
- Remote personal server configuration
- Authorization
- Zapier account and authorization for each connected app
- Status
- documented
An action can change an external app; actions are not separate servers and task limits apply.
53. n8n Instance-level
Expose permitted workflows and supported workflow-building and data-table tools.
Example task: Inspect a sandbox workflow before an authorized test run.
- Connection
- Remote Cloud or self-hosted instance endpoint
- Authorization
- OAuth or instance API key
- Status
- documented
Enable MCP and workflow access deliberately; running a workflow can trigger its published external actions.
54. Make
Expose scenarios and supported management tools.
Example task: Inspect an approved scenario before a test execution.
- Connection
- Remote Streamable HTTP
- Authorization
- OAuth or regional MCP token
- Status
- documented
Scenario execution consumes credits; management access and timeouts depend on the configuration.
Payments and market data
These entries describe software access, not investment choices or financial recommendations. Use sandbox accounts for payment records and paper accounts for trading integrations. Market-data timestamps, licensing and plan access need their own review. No entry authorizes a real payment or order.
55. Stripe
Use supported account, payment and subscription context or operations.
Example task: Inspect a sandbox subscription's status.
- Connection
- Remote HTTP
- Authorization
- Stripe OAuth or agent API key
- Status
- documented
Live and sandbox permissions are separate; payment-changing actions require explicit approval.
56. PayPal
Use supported order, invoice and subscription tools.
Example task: Inspect an invoice in the sandbox.
- Connection
- Remote server or local stdio
- Authorization
- PayPal OAuth/access token for the chosen setup
- Status
- documented
Select the sandbox endpoint or SANDBOX environment explicitly; do not run live money-changing actions.
57. CoinGecko
Retrieve cryptocurrency market and historical data.
Example task: Compare historical data while retaining timestamps.
- Connection
- Remote HTTP
- Authorization
- Public beta access without a key; extended access has separate terms
- Status
- beta
Beta limits may change; data is not a price guarantee or a trading recommendation.
58. Dune
Use blockchain datasets, SQL queries and saved analysis tools.
Example task: Inspect the SQL behind an on-chain metric.
- Connection
- Remote HTTP
- Authorization
- Dune OAuth or API key
- Status
- documented
Queries consume credits; changing a saved query is not a blockchain transaction.
59. Alpaca
Retrieve market/account context and use supported order tools.
Example task: Inspect a paper account's positions.
- Connection
- Local server
- Authorization
- Alpaca API key and secret
- Status
- documented
Order tools can trade when configured for live access. Keep paper mode explicit; this catalogue gives no trading advice.
60. Massive (formerly Polygon.io)
Retrieve supported US market data through the hosted MCP server.
Example task: Inspect historical market data with its timestamps.
- Connection
- Remote HTTP
- Authorization
- Massive account OAuth
- Status
- documented
Dataset access and latency depend on the plan; this is the market-data company, not the Polygon blockchain.
Official documentation · Official source 2 · Official source 3
First-connection checklist
- Confirm the official server identity and use the current provider setup page.
- Check transport and authentication support in your specific client.
- Select a test account or narrow project; grant only the required access.
- Inspect the tool list and separate reads from writes, sends, publishes and paid executions.
- Ask for one read operation with an identifiable source or record. Compare the response with that source.
- Keep retrieved content as data. Do not let instructions inside a page or document expand the approved task.
- Know how to revoke the connection or token before enabling further tools.
For a repository task, follow the GitHub MCP guide. For workspace pages, use the Notion MCP guide. The Cloudflare documentation MCP guide covers public documentation lookup; it is different from the Cloudflare API server in entry 26, which can execute account operations.
Questions about MCP servers
Do I need to install all 60 servers?
No. Start with the server that can access the input for one task. More connections introduce additional permissions and tool choices to review. Add a second connection only when you can explain the operation it contributes.
Does a remote server work in every AI client?
No. Check the client’s supported transport and authentication method. An HTTP endpoint, OAuth support or a connector listed in one application does not establish support in another application or account tier.
Are local MCP servers free and private?
A local process can call external paid APIs and send data to them. Review its code, configuration and provider terms. File and browser tools also inherit access from their execution environment, so local placement alone is not a privacy boundary.
Can I use Google Workspace’s developer MCP servers now?
Google’s reviewed setup guide requires Developer Preview Programme membership, a Cloud project, enabled APIs and MCP services, and OAuth configuration. Verify your eligibility before setup. This requirement is separate from an AI app’s existing Gmail or Drive connector.
Which server is best for my first connection?
Choose one that matches an input you are allowed to share and an operation you can verify. Public documentation lookup avoids account writes; a scoped test repository may be more useful for a coding task. There is no benchmark ranking in this catalogue.
What should I do if the listed tools or login options have changed?
Open the official source attached to the entry and check the current setup instructions. Revisit permissions when reconnecting. The review date describes documentary research, not a guarantee about the endpoint’s live state.
Make a one-server shortlist
Download the CSV, keep the row that matches your task and note the access you will grant. Then use the relevant setup guide and inspect one result before enabling writes or adding more connections.