MCP·ExplorerIndex 001

MCP servers worth giving an agent.

Not another scraped directory. A sharp index of agent-ready servers with transport, install command, provenance, and the actual reason each one earns a slot.

More
on the way
10
official vendor
13
categories
2026
last reviewed

The index

001–015
01GitHub
official·Code·remote

Read repos, manage issues & PRs, search code. The reference integration for engineering agents.

remote https://api.githubcopilot.com/mcp/
02Linear
official·Project mgmt·sse

Create, triage, and update issues and projects straight from the agent loop.

remote https://mcp.linear.app/sse
03Notion
official·Docs·stdio

Search, read, and write Notion pages and databases — the team knowledge layer.

npx -y @notionhq/notion-mcp-server
04Stripe
official·Payments·stdio

Query customers, create payment links, inspect charges. Money tools with real consequences.

npx -y @stripe/mcp
05Supabase
official·Database·stdio

Manage projects, run SQL, inspect schemas across your Supabase Postgres.

npx -y @supabase/mcp-server-supabase
06Sentry
official·Monitoring·remote

Pull issues, traces, and release health so agents can debug from real telemetry.

remote https://mcp.sentry.dev/mcp
07Cloudflare
official·Infra·remote

Workers, DNS, R2, and observability — operate your edge from the agent.

remote https://observability.mcp.cloudflare.com/sse
08Playwright
official·Browser·stdio

Drive a real browser with structured accessibility snapshots, not pixel-guessing. By Microsoft.

npx -y @playwright/mcp
09Figma
official·Design·sse

Pull design context — frames, variables, components — into code-gen via Dev Mode.

remote http://127.0.0.1:3845/sse
10Neon
official·Database·stdio

Spin up serverless Postgres branches, run migrations, query — all agent-driven.

npx -y @neondatabase/mcp-server-neon
11Filesystem
reference·Files·stdio

Scoped, permissioned file access — the bedrock tool for local coding agents.

npx -y @modelcontextprotocol/server-filesystem
12Firecrawl
community·Web scraping·stdio

Crawl, scrape, and extract clean markdown from any site with JS rendering handled.

npx -y firecrawl-mcp
13Context7
community·Docs·stdio

Up-to-date, version-correct library docs injected on demand. Kills stale-API hallucination.

npx -y @upstash/context7-mcp
14Exa
community·Search·stdio

Neural web search and content retrieval built for agents, not humans skimming SERPs.

npx -y exa-mcp-server
15ApexLinkedIn
LeadShark·LinkedIn·remote

Governed LinkedIn actions for agents: discover, research, comment, connect, message — paced and logged.

remote https://apex.leadshark.io/mcp

How we pick

METHOD
A

Hand-picked, not scraped

Every entry is here on merit — real maintenance, real docs, a tool surface worth trusting. No ranking by star count.

B

Official sources first

We prefer vendor-owned and Anthropic reference servers, and label provenance plainly so you know what you're installing.

C

Capability with consequences

Read vs write vs destructive matters. The best servers make auth, pacing, approval, and auditability part of the product surface.


For agents & operators

API

The full index is machine-readable — search the API, fetch JSON manifests, or read the markdown mirrors. Every record carries provenance and a last-checked timestamp.