MCP server directory

Security-first, self-hosted AI brain for email. Synchronizes IMAP mailboxes into user-owned PostgreSQL and exposes tools for listing, searching, reading, and cited questions across the archive. Supports local chat and embedding models through OpenAI-compatible endpoints; each deployment exposes its own Streamable HTTP `/mcp` endpoint.

A Model Context Protocol (MCP) server that gives AI assistants (VS Code Copilot, Cursor, Claude, Gemini, Antigravity) the AccelByte Extend SDK as additional context — so they can answer questions about the SDK and generate correct Extend SDK code. - Search Extend SDK symbols (functions and models) by name, tags, or description (fuzzy matching) - Describe specific symbols — parameters, fields, imports, examples, return types, required permissions - Scaffold a new Extend app from a template repository via the create-extend-app prompt

3 tools · Describe specific symbols, Scaffold a new Extend app

Persistent memory for Claude Code, Cursor and Codex where a replaced fact is retired at write time, so what an agent reads back is the current answer rather than the one that was true in March. Facts are typed and carry provenance and a review trail; nothing is deleted, it stops being current. Local SQLite, queries answered from a local replica and never sent to a server, no API keys. `npx -y @dat999zx/knowl serve`

AI-driven blueprint automation for Unreal Engine 5.1-5.8: a C++ bridge plugin (59 editor UFUNCTIONs) exposes node-level graph authoring that engine Python cannot reach - create/wire/compile blueprints, author materials and UMG widgets, with read-back verification on every write and PIE runtime state checks. Ships prebuilt DLLs for 5.1 and 5.8 (zero compilation); MCP endpoint with 23 tools.

Curated access to a self-hosted [Mealie](https://mealie.io) recipe manager: 52 tools for recipes, website/photo imports, meal plans and shopping lists, with confirmation tokens on destructive operations and a read-only mode.

Official MCP server for the Algenta decision engine. Dataset discovery, governed exact queries, utility models, decision memory, and agent runs under tool governance profiles (observe/govern/execute/full), with approvals and execution receipts. `pip install algenta-mcp`

> Institutional-grade manager tearsheet generation for alternative investment research — powered by AI agents, served via MCP. AltBots Tearsheet is an end-to-end AI pipeline that generates structured, institutional-quality research tearsheets on alternative investment managers — hedge funds, private equity firms, and family offices — from publicly available data and proprietary intelligence signals. Built by a former Managing Director with 20+ years of institutional buy-side experience (2,000+ manager reviews, $9B+ deployed), this system encodes the actual diligence framework used at multi-billion-dollar investment offices into a fully automated, agent-driven research workflow.

4 tools · generate_manager_tearsheet, get_manager_coverage

AI SEO audit and fix tools: score any site 0-100 against 50+ AI crawlers (ChatGPT, Claude, Perplexity, Google AI), validate JSON-LD and robots.txt, compare human vs AI-crawler views, and generate llms.txt, robots.txt, and JSON-LD schema. Install via `npx -y crawl-readiness-mcp`.

Outbound-only remote shell, detached long-running jobs, and temporary file courier for AI agents (https://aicommander.dev). Hosted Streamable HTTP; no exposed SSH/ports/VPN. Install via npx -y @aicommander/mcp.

Local code search for AI coding agents: hybrid keyword + semantic search with SQL relevance ranking over a plain-file index. No accounts, keys, or server.

A safety gate and explainer for AI coding agents that run shell commands. `check_command_safety` returns a danger/caution/none verdict for a proposed command — catching `rm -rf /`, `curl | sudo bash`, `dd`/`mkfs`/`shred` to a device, `chmod -R 777 /`, fork bombs, `git push --force` and CI `${{ }}` script-injection — so an agent can guard a command before executing it. `explain_command` gives a token-by-token breakdown of every flag, pipe, redirect and subshell. Fully offline, zero dependencies, nothing leaves the machine. `npx -y cmdxray mcp`.

PlayStation Network gaming history: playtime, trophies and purchased library, for grounding game recommendations in what you actually played.

Read and drive a self-hosted [Woodpecker CI](https://woodpecker-ci.org) instance: find the failed pipeline, read the failing step's log as text (the API returns base64 chunks), then restart, cancel or approve it, and manage secrets, registries, crons, agents and the server queue. 71 tools, or a curated eight via `WOODPECKER_ALLOW_TOOLS=essential`; irreversible operations need a confirmation token and agent tokens are redacted. `npx -y @ni-c/woodpecker-ci-mcp`

Turns a company domain into its registered legal entity: legal name, company number, jurisdiction, status, LEI, and VAT number.

Fast typed judgments from Jev (TypeSafe System One): the official choice/score/noul question types with calibrated probabilities, plus A/B compare, claim-vs-evidence verify and batch classify — sub-second, sub-cent decisions for triage, routing, risk scoring and gating in coding agents. One-command client install: `uvx jev-mcp-server install claude-code`.

Run Jev AI typed decisions (choice, score, noul) on jevx.org and get calibrated probabilities on every option. Zero-shot: criteria live in the request, no training or fine-tune to change the label set. `npx jevx-mcp`.

MCP server for Appcircle: exposes Build, Signing Identities, Testing Distribution, Enterprise App Store, Publish to Stores, and Reporting tools to any MCP-capable client (Claude Desktop, Cursor, VS Code, etc.). The Appcircle MCP Server acts as the bridge between AI tools and Appcircle; thus, AI agents, assistants and chatbots to safely access and interact with Appcircle resources through structured, governed, and task-level tools. - CI/CD and Workflow Intelligence: Monitor pipeline runs, track release status, and get insights into your mobile CI/CD workflows. - Configuration and Environment Insights: Query build configurations and signing setup to understand how a project is configured and where issues may originate. - Reporting and Operational Insights: Generate summaries of CI stability, recurring issues, pipeline performance, and overall CI/CD health.

6 tools · get_build_insights_report, get_build_status

Verify blockchain-anchored timestamp proofs locally: recompute a file's SHA-256, walk the Merkle path, and check the anchor on Polygon or Bitcoin. Verification makes no network call and needs no account or API key, and never asks the issuing service whether its own proof is good; an optional tool anchors a new hash through ProofLedger. `pip install "verify-proof[mcp]"`.

Local image-workflow MCP for AI coding agents: generate and edit images via your own OpenAI-compatible or Gemini image APIs and save results into the project directory, with optional model fallback. `npx -y gen-image-mcp`

Independent judgment on DeFi vaults for agents: a 16-factor Atlas Score with full factor breakdown for 500+ vaults (Morpho, Aave, Pendle, Yearn, Beefy, Euler on Ethereum, Base, Arbitrum, Optimism, BNB), open blowup alerts, and route survival: whether a deposit can be exited at size today, measured by a real USDC round trip rather than an advertised liquidity figure. A `NO_ROUTE` answer comes back as `blocking: true` so the agent is told not to deposit. Read-only, no key, no custody; scores published daily to a public history repo since June 2026. `npx -y @atlasyield/mcp`

Local-first AI media workbench with a Canvas, short-form timeline editor, shared Assets, and 29 MCP tools for agent-controlled projects.

The solve brain for CAPTCHA workflows: identifies CAPTCHA type and task parameters, serves CapMonster Cloud docs, creates solving tasks, and retrieves results via the CapMonster Cloud API. In the official MCP Registry as `io.github.CapMonsterCloud/capmonster-mcp-captcha-solver`. `uvx capmonster-mcp` or `npx capmonster-mcp`

Compiler-as-MCP-server for the Hirð language (typed actors and effect rows on the BEAM): type inference, definition lookup, effect-row explanation, IR fragments, actor protocol and supervision graphs, and token-budget-aware symbol summaries, so agents query the compiler instead of reading source. Errors come back structured, so an agent can self-correct from tool output alone. `cargo install --git https://github.com/no-materials/hird hird-mcp`

OpenStreetMap for travel planning: geocoding, walking/driving/cycling routes with correct FOSSGIS profiles, travel-time matrices, route optimization, isochrones and POI search. No API key required; client-side rate limiting keeps it inside the public services' usage policies. `npx -y osm-mcp`

Pushes enriched lead rows into an Instantly or Smartlead campaign with ICP score gating, deduplication, and a zero write dry run.

Certified SEC EDGAR fact retrieval with cryptographic filing provenance, zero LLM hallucination, valuation packs, and deterministic CAGR calculation. Remote streamable-HTTP at https://finsec.caleralabs.com/mcp.

One todo list and backlog every coding agent can write to (Claude Code, Codex, Cursor and any MCP host), filed per project from the git remote, with claims so two agents don't pick up the same item and a local web dashboard. Local-first, no account; optional self-hosting.

Detects whether a company only declares AI, actually deploys AI tooling on its site, or charges money for AI, returning a four-level AI maturity tier with the evidence behind the verdict.

Profile any MCP server into a reusable, read-only skill that captures how you actually use it — your workflows, the identifier hops that chain one call into the next, and each tool's safety class. Exploration runs behind a default-deny gate that provably never writes (a `SELECT` runs; a `DROP` on the same tool is refused by inspecting the statement). One static binary. `npx @ieranama/discomcp`

Autonomous AI-only labor marketplace protocol with atomic escrow and weighted anti-Sybil reputation.

EU AI Act measurement corpus — 417 frozen provisions (the Act's 113 Articles at provision-level granularity) with risk classification, a 42-point audit, Article 11 documentation drafts, and a penalty calculator. Measurement, not certification. Install: `pip install eu_ai_act_compliance_mcp`.

📇 ☁️ Enable AI Agents to fix build failures from CircleCI.

6 tools · config_helper, download_usage_api_data

Resolves a company domain to its Bluesky account with exact follower, following, and post counts.

MCP server for the Hyperiux Vault effect registry: three read-only tools to browse/search 150+ React/Next.js animation and interaction effects (50+ free, 100+ Pro), fetch full detail (dependencies, changelog, install command, and source for effects you're entitled to), and list categories. Install via NPM: `npx -y hyperiux-mcp-server`