Development MCP Servers

Publish and manage HTML5 games on Playgama from a coding agent: game form, builds, covers, in-app catalog, leaderboards, sandbox link.

Live failure intelligence for AI agents. Check whether other agents are hitting the same tool failure and see which recovery actions actually worked before retrying.

One folder of Markdown holding a reviewed codebase ontology and a cited wiki. 38 typed tools, local-only, every write is a Git diff.

& SQL injection, exfil sinks). Free MCP server + paid x402 API ecurity scanner for n8n workflow exports: 18 static-analysis rules mapped to OWASP Agentic AI (webhook auth, hardcoded secrets, SSRF/IMDS, command (0.50 USDC/scan, settles on Base).

MCP server for Azure Service Bus: peek and search message bodies across queues, topics and dead-letter queues, then resubmit, send, or purge. Read-only by default, with sending and destructive operations behind separate opt-in flags.

Paid MCP tools for evidence capture and preflight verification before autonomous agent actions, settled via x402 on Base.

Audit Python and npm dependency licenses before you ship. CLEAN / REVIEW / BLOCK verdicts, context-aware, runs fully locally over stdio.

Find what slows your website down. Use Claude Code, Cursor or ChatGPT to help fix it. Then check that your site is faster.

Run commands on a machine you cannot SSH into. A step goes out over git, a file share or an HTTPS relay, and the whole run comes back as a log with every line timestamped in UTC, whether it passed or failed. For air-gapped, bastion-only and change-controlled estates: no tunnel, no proxy, nothing held open.

Check i18n locale files from your editor or agent: missing keys, dropped placeholders, collapsed plurals — with validator tools that make zero model calls and need no API key. Optional semantic review and translation with your own LLM key. JSON, Flutter .arb, Apple .xcstrings. Apache-2.0.

Independent evidence for how reliably AI agents select MCP tools, with frozen benchmark results, preserved semantic misses, multi-batch reliability evidence, and independent MCP evaluation.

Give your AI hands. One URL. Type a job. Get a receipt.

Load testing and synthetic monitoring. Test your site with Playwright, Browser Bots, or Protocol Bots. Use the MCP to build and play test scripts, configure load tests, analyze test results, and set up monitors.

X12 EDI developer MCP for docs, synthetic fixtures, governed connections, and profile-gated sandbox/production API workflows (850/810/856/837P).

The complete 4-in-1 MCP toolkit for Claude Desktop and Cursor: AST token compressor, config auto-repair, semantic git & universal schema compiler.

Swift code reviews and local Apple source/guide retrieval for Claude and other MCP clients. Optional Xcode simulator tools build, test, launch and preview apps.

Open source MCP integration layer that connects any API to any AI model with pre built connectors, zero credential storage, and a hosted Hub. Connect any API to any AI model with pre built connectors, a hosted Hub, and zero credential storage on Corsair's side.

Inspect the certificate and TLS protocol versions served by a public endpoint. Checks hostname coverage and expiry; does not verify trust chains, revocation, or cipher suites.

Local code search for AI coding agents — hybrid keyword + semantic search with SQL relevance ranking over a plain-file index. CLI and MCP server, no accounts or keys.

Metered coordination for autonomous agent swarms: locks, budgeted sub-keys, shared file storage, messaging, and webhooks — billed per operation via x402 (USDC), no subscription.

Offensive360 MCP server: run static application security (SAST) scans across 60+ languages and read findings with secure code fixes from Claude, Cursor and other MCP clients. Works with the Offensive3

Remote MCP for fetch, GitHub shape, weather, docs.

Vocemo is an on-device AI toolkit for macOS. It types what you say into any app, transcribes recordings with speaker labels and captions, reads text aloud, answers questions about what is on screen, converts documents, and translates. Every model runs on your own Mac, so no file, recording or word is ever uploaded. The installer is 5.5 MB because models download only for the features you use, and shut down when idle. One subscription covers every feature on up to three Macs, with a seven day free trial.

Read-only checks over a repository's own guardrails, run over MCP or from the command line, that fail a check which is present and has stopped refusing anything. Every answer is appended to a local, append-only ledger. No network calls, no telemetry, no writes to the repository being checked.

AI-powered testing platform: generate tests, mocks, and sandboxes from real API traffic.

Query live, sourced AI-model pricing and capability data — compare models and route a task to the cheapest one that clears a capability bar. Covers image, language, video, audio, and coding/science/agentic/GDPval benchmarks.

Give your coding agents a shared task board and knowledge base, so the plan survives between sessions and across agents.

Remote MCP and A2A infrastructure for autonomous agents: deterministic repair utilities, trust signals, intent discovery, and HTTP x402 payments on Base.

Transform your database or API into MCP tools in seconds

Shiplight's browser MCP server lets coding agents inspect web pages, perform browser actions, and capture screenshots for testing and debugging.