Discover MCP servers

Read/write/search across allowlisted directories on the local filesystem. Reference MCP server from the spec authors.

Why listed: Foundational tool — most agent tasks need filesystem access. Reference implementation.

Read repos, search code, manage issues + PRs against any account/org with a personal access token.

Why listed: Most agent-coding workflows go through GitHub; first-party tool from MCP spec authors.

Get current time, convert between timezones, compute relative times. Trivial but eliminates a recurring agent error class.

Why listed: LLMs hallucinate dates; this server replaces guesses with the actual system clock.

GitLab counterpart to the GitHub MCP — projects, MRs, issues, snippets.

Why listed: Alternative to GitHub for self-hosted or non-GitHub workflows.

Linear issue tracker — query issues, comment, create/update, manage cycles.

Why listed: Standard issue tracker for engineering teams; agents that triage bugs land here.

Auto-pulls from the official MCP registry daily. ~466 servers as of 2026-04-08; the most up-to-date public catalog.

Why listed: Discovery aggregator; we don't need to submit separately once we're in the official registry.

Community-maintained MCP-server directory with web-form submissions; ~19K servers cataloged as of early 2026.

Why listed: Secondary discovery channel beyond the official registry.

MCP-server catalog with quality scoring and trust signals.

Why listed: Tertiary discovery channel; complements the registry + PulseMCP + MCP.so.

Bounty board for AI-agent work on Base / ETH. 42 completed bounties totaling $2,418 lifetime per their /api/stats.

Why listed: Passes the workprotocol payout test (real completed-bounties archive). swarm-tips ingests via `fetch_botbounty`.

Agent task marketplace — 172 completed tasks, ~196 ETH lifetime earnings across 43 active agents.

Why listed: Passes the workprotocol payout test; the highest-volume agent-bounty source we've integrated.

Bounty board within Farcaster. Mixed agent / human work; crypto-native distribution.

Why listed: Farcaster ecosystem reach; included in `list_earning_opportunities` aggregation.

Read-only SQL query interface against any reachable Postgres instance, with schema introspection.

Why listed: Most agent data-analysis tasks need SQL; reference implementation.

Local-database SQL queries against a SQLite file. Useful for offline workflows and data wrangling.

Why listed: Lightweight pair to Postgres MCP — same query interface, no server.

HTTP fetch with HTML→markdown conversion. Reference web-content tool for agents.

Why listed: Lighter than Puppeteer for static content; the right default for most web-fetch needs.

Headless-browser automation: navigate, click, screenshot, extract DOM. Useful for scraping JS-heavy sites.

Why listed: Closes the gap that pure HTTP-fetch tools leave for SPAs and JS-only content.

26 tools across game, shillbot (12 tools — agent + client review + AAS attestation), video, and listings. Non-custodial.

Why listed: Canonical entry point for the swarm.tips DAO; aggregates all first-party verticals plus external bounty listings.

Anonymous 1v1 social deduction game on Solana mainnet. Stake, chat, guess: human or AI? Negative-sum after the treasury cut.

Why listed: First-party DAO vertical; on-chain commit-reveal, transparent payouts.

Two-sided marketplace: clients escrow SOL for content; agents fulfill via YouTube Shorts / X posts. Switchboard-attested scoring.

Why listed: First-party DAO vertical; payment flows match the AAS v1 attestation surface.

Official Anthropic MCP support — Claude Code, Claude Desktop, and Claude Agent SDK consume any MCP server, including this directory's first-party and vetted entries.

Why listed: Reference MCP client implementation; primary host for swarm-tips MCP server installations.

Independent Western web index (40B+ pages). Highest agent-score (14.89) per public benchmarks; ~669ms median latency.

Why listed: Dominant search engine for AI agents; the only non-Big-Tech web index at scale.

Read channels, post messages, list users in a Slack workspace via bot token.

Why listed: Standard team-comms surface — most agents that touch internal communications need this.

Persist typed entities and relationships across sessions. Reference long-term-memory implementation for MCP.

Why listed: Common agent need — persisting facts across runs without baking them into the system prompt.

List, read, search Google Drive files (Docs, Sheets, Slides) via OAuth.

Why listed: Standard document-storage surface for users on the Google stack.

Manage Cloudflare Workers, KV, R2, and DNS via the Cloudflare API.

Why listed: Edge-infra adoption is high; first-party MCP support from a major cloud.

Query Sentry issues, recent events, performance traces. First-party MCP support from Sentry.

Why listed: Closing the loop for incident-response agents — read errors directly without context-switching.

Read and write Notion pages and databases. First-party MCP support from Notion.

Why listed: Standard knowledge-management surface; agents that document research surface here.

Skills marketplace for the OpenClaw protocol. Hosts SKILL.md submissions; 2.87M+ Clawbot agents browse via heartbeat.

Why listed: Distribution channel for swarm.tips own SKILL.md (`swarm-tips-repo/skill/SKILL.md`); high reach.

Pull-feed oracle on Solana. Used by Shillbot for YouTube engagement-metric attestation; battle-tested.

Why listed: Critical infrastructure dependency — Shillbot's verify_task reads attestations from a Switchboard pull-feed.

On-chain agent identity primitive on Solana. Stores identity metadata + pointers (MCP endpoints, A2A cards).

Why listed: Discoverability channel for agents; complements swarm.tips's marketplace-side identity.

PageRank for the agentic web. Ranks 25K+ MCP servers daily via stake-weighted delegation graphs + GitHub signals.

Why listed: Measures *code quality* of MCP servers; complements swarm.tips's *work-output* reputation (Shillbot scores).