Development MCP Servers

Open-source proxy that sits in front of any MCP server and gates tool calls by whether their effect can be undone: reads pass, reversible changes are rolled back if the run fails, irreversible calls wait for a human to approve the exact request.

Translates MCP tool errors into next actions — Solo $9, Fleet $19 /mo + $0.002/check

Compatibility-typing for ingested context — Solo $9, Fleet $19 /mo + $0.005/check

Evidence verification and permission-based environment diagnostics for AI agents.

MCP server for Claude, Cursor and other clients: it asks the intake questions and renders 22 SOC 2 policies from Apache-2.0 templates.

AI personas browse your web app like real users, report usability bugs and return copy-paste fix prompts for Claude Code and Cursor.

Zero-dependency AI coding agent context baton, 80-90% AST-like code skeletonizer with lossless on-demand MCP rehydration (vicem_rehydrate), and cross-platform desktop utility GUI (npx vicem-cli gui) across 12 coding agents (Claude Code, Codex CLI, Gemini CLI, Cursor, Windsurf, OpenCode).

Hosted AI agents for coding and automation, available as remote MCP servers (Streamable HTTP, bearer token) and through a REST API for Zapier and n8n. Monthly plans from $10.

APIClaw is a flat-rate, OpenAI-compatible AI API gateway. Its remote MCP server lets an agent discover and call LLM access, plan catalog and usage endpoints through MCP: one key reaches Claude, GPT, Gemini, DeepSeek, Qwen, Kimi and 100+ models from $19/mo, with unlimited tokens per request and a daily request quota resetting at 00:00 UTC.

Kerno runs locally inside your coding agent's loop and tests every change against your running app and real dependencies, catching regressions and security issues before you merge.

Build & publish full HTML5 websites with working forms, generate master prompts, and create textured 3D models streamed directly to Roblox Studio via remote Streamable HTTP MCP server.

Hosted MCP server (OAuth) for the Safeguard software supply chain security platform: look up projects, SBOMs, findings and vulnerabilities with EPSS and CISA KEV context.

x402 machine-payable API edge: pay-per-call utilities, market data, audits and agent tools over MCP. Base USDC, EIP-3009 gasless.

An MCP server that gives AI assistants (Claude, Cursor, etc.) direct access to Jamf official documentation. Ask Jamf-related questions and get answers based on the latest docs from learn.jamf.com.

Check whether GPTBot, ClaudeBot and other AI crawlers can access your site.

Free, open-source Mac app and MCP server that lets AI agents take screenshots, tap, swipe, type and launch apps on a real iPhone, with owner-approved clients, activity logs and pause controls.

Write once. Let your agent publish. Open-source, local-first CLI + MCP server that publishes one Markdown file to dev.to, Hashnode, WordPress, Ghost, Medium, Substack, Pages & X Articles. Free / open source (MIT), self-hosted. Website: https://openpub.princepal.dev/ Docs: https://openpub.princepal.dev/docs

Burs-IA offers MCP agent capability discovery and governed technical interoperability checks. No security certification; ONDA external messaging is not yet available.

Capture website screenshots from AI agents via the ScreenshotRun API — PNG, JPEG, WebP with viewport, device emulation and element targeting

Scans an MCP server for security gaps before your agent connects. Free tier, no signup.

Hosted MCP server with OAuth sign-in, and plugins for Claude Code, Codex and Cursor: coding agents read each story's acceptance criteria, designs and tests from Stride and update it as they work.

Build, check and fix n8n workflows from your AI assistant: 1,100+ n8n node definitions, curated templates and a validator.

Scores whether a page reads as AI-generated and names the tells to fix.

Secure OAuth remote MCP control of explicitly paired computers: files, terminal, browser and desktop automation. Device operations can modify state.

40x fewer tokens while reading the RFC corpus, with obsoleted specs flagged

Logs your AI agent into websites, catches verification codes, and asks you before anything risky.

Turn a repeatable task into a self-correcting AI agent loop; a deterministic gate, not the model, decides when it's done.

LadDoc → SimaticML 云服务 MCP：严格校验梯形图文档（LadDoc IR），确定性生成可被 TIA Portal V19 直接导入的程序块（FC/FB/OB）。

Calibration Court MCP from ScriptMasterLabs: model-calibration leaderboard, per-model reports, and verification of a hash-chained calibration ledger. Paid recalibration endpoints settle in USDC on Base via x402 ($0.01 / $0.05).

Canonical x402 Gateway MCP from ScriptMasterLabs. Free discovery surface: find the x402 Gateway contract, inspect gateway status, and read the canonical payment contract (Base mainnet, USDC, payTo and header rules) before an agent pays anything. Paid capabilities are resolved per-target from fresh payment requirements.