What works — not just what exists.
Production-grade Model Context Protocol servers, vetted against a public reliability score. The directory CTOs and engineering teams trust — not another junk drawer of broken links.
Hand-picked, production-ready.
Servers that pass every reliability gate and have proven themselves in real deployments.
Secure file operations with configurable access controls.
Issues, PRs, code search, releases — the full GitHub surface.
Read-only SQL queries against a Postgres database.
Send, read, and search Slack messages from your agent.
Privacy-respecting web and news search.
Fetch any URL, auto-convert to markdown.
Find what you need by domain.
Browse servers organized by workflow — not by whichever category sounded clever.
MCP servers that give agents web, documentation, and code search. Essential for grounding responses in up-to-date facts.
Servers that fetch, parse, and structure data from the web. Used for extraction, research, and ingestion pipelines.
Integrations with Slack, Discord, email, and collaboration tools so agents can read, send, and manage messages.
Task management, note-taking, calendars, and personal-automation servers — the MCPs your agent uses to get real work done.
Code execution, linting, test running, package registries, and build tooling. MCPs that let agents ship code, not just describe it.
Postgres, MySQL, MongoDB, Redis, and vector store connectors. Read-only and read-write MCPs for structured data workloads.
AWS, GCP, Azure, Vercel, Cloudflare, and other cloud provider integrations. Provision, inspect, and operate infrastructure via agents.
Local and remote file system access. Read, write, search, and manage files with controlled scope.
S3, R2, Google Drive, Dropbox, and object storage MCPs. Upload, download, and organize assets from agent workflows.
Git, GitHub, GitLab, Bitbucket. Issues, PRs, branches, code review — everything your engineering agent needs from source control.
Model gateways, embeddings, fine-tuning, and evaluation tools. MCPs to compose agents that use other models as tools.
CI/CD, containers, Kubernetes, observability. MCPs that let agents diagnose incidents, roll out fixes, and monitor systems.
BI tools, event streams, and metric dashboards. Query product and business data without leaving the agent.
Secret scanning, vulnerability feeds, SAST tooling, and auth integrations. The guardrail layer for agent-driven operations.
Servers that don't neatly fit another category. Specialized integrations, experimental tools, and niche use cases.
New on MCPfinder.
The most recently added servers, each already screened against our reliability bar.
Manage containers, images, and compose stacks.
Read and write Notion pages and databases.
Create and update Linear issues from your agent.
Read AWS resources — EC2, S3, IAM, and CloudWatch.
Time-zone aware clock, conversions, and scheduling helpers.
Reference server showcasing every MCP primitive.
Built for the CTO who has to ship this.
Other directories tell you what exists. MCPfinder tells you what works in production — and gives you the receipts.
Public 0–100 score from daily automated checks. If a server goes stale, it gets flagged — not hidden.
Install commands per package manager, version history, usage examples, maintainer signals, alternatives.
Filter by language, runtime, license, last release. Decide in seconds what other directories make you scroll for.