Development MCP Servers

Free advisory policy checks and JSON data cleaning for autonomous AI agents. Integrators enforce ALLOW/BLOCK; no payment execution or wallet control. Python, TypeScript and remote MCP examples.

Query-aware context compression for coding agents. Trims tool output, logs and diffs before they reach the model.

Live npm and UNPKG data for AI agents: package resolution, project audits, import maps, and malware/CVE checks.

Hands work to another coding agent over ACP: Claude Code, Codex, OpenCode, Gemini CLI and others

Deterministic offline repair of malformed JSON from language models: code fences, prose noise, trailing commas, smart quotes, unquoted keys, bare-word values, plus required-key checks. MCP server + skill, Apache-2.0. Fix desk for broken agent/MCP configs: https://skynet.website/services/

- [toolmenu](https://github.com/niksa90/toolmenu) - Deterministic MCP server tool-list analysis that catches breaking changes, confusing tool descriptions and prompt-cache disruptions.

Remote MCP server for Actera, the guardrail layer for AI agents. From Claude, review what your agents did, see what is waiting for approval, and draft, test and publish policies that allow, block or require approval for agent actions. OAuth sign-in; works with claude.ai, Claude Desktop and Claude Code. Endpoint: https://mcp.actera.dev/mcp

Self-hosted key gateway for AI agents: the agent requests access over MCP, the owner approves it in Telegram or the web UI, and the agent gets a temporary pass limited by time and number of calls instead of the API key. Read-only mode, per-field data filter and audit log.

Free hosted jq MCP server and REST API. Run real jq on inline JSON, public HTTPS URLs, or reusable 15-minute uploads. No install or signup.

Token due-diligence for AI agents: letter-grade verdicts, mint activity, settlement safety, token and wallet profiles. Payable per call over x402 on Base; free developer key tier available.

Public keyless Streamable HTTP MCP server to search curated SaaS and AI tools, browse categories and tags, submit your own product, and verify backlink badges.

Remote MCP server to track brand visibility, citations, prompts, competitors, and missing citation sources across ChatGPT, Perplexity, Gemini, and AI Overviews.

Remote MCP server to generate AI slide decks and pitch decks from a prompt and export native PowerPoint (.pptx) and PDF presentations from Claude & Cursor.

Remote MCP server for domain authority & backlink research, SERP & keyword discovery, AI search visibility checks, publisher site approvals, and link ordering.

Expose 248 text tools to compatible clients through String Utility Belt’s MCP server.

One API for video, image, and audio models. Remote hosted MCP over Streamable HTTP. Docs: https://kinovi.ai/docs/mcp

OpenAny (openany.si) is the marketplace for agents: one key for 168+ AI models and 3,200+ APIs. No subscriptions, no per-provider signups. Top up once, pay per call. Remote MCP.

Dynamic MCP server that auto-generates tools from OpenAPI/Swagger specs. Published on PyPI (nife-restapi-mcp-server).

Dynamic MCP server that auto-generates tools from any GraphQL schema. Published on PyPI (graphql-mcp-server) and npm (@nifelabs/mcp-server).

Paste an exact error string an AI agent hit, get a checked fix and a verify step from the Agent Field Guide. Free, open, read-only MCP. As is.

SSH, SFTP, databases and monitoring for your own Linux servers from any MCP client: run commands and jobs, edit files, query databases, read metrics and alerts, without handing the agent an SSH key. Hosted at https://termal.in/mcp, OAuth 2.1 sign-in, per-server command policy, audit log. 23 tools.

TypeScript MCP server that lets Claude work with a store REST API (inventory, orders, sales). Writes are guarded by dry-run, runs over stdio or Streamable HTTP, includes a demo API and Vitest tests.

Reed-Solomon assistant: encode, decode and correct data, syndromes, GF(256) math. Written in Rust.

Documentation MCP for the Manifold-BT backtesting engine. Gives coding agents the DSL rules, indicator and config signatures, worked examples and error fixes. Read-only, runs no backtests.

Use any AI model from the app you already work in: ask GPT from Claude, Claude from ChatGPT, or Gemini and Grok from Cursor or Codex. Name one model or several. Every answer says what paid for it: your ChatGPT plan (local client), your own API or OpenRouter keys, or 10 free requests. Hosted at https://polydev.ai/mcp with OAuth, or run locally with npx polydev-ai.

OpenHAL is a universal driver specification that lets AI agents safely discover, set up and control lab and manufacturing equipment through the same read and write commands. Each device has a file of limits a person confirms, OpenHAL checks every command against that file, and actions that can't be undone are held for a person's approval. OpenHAL has 36 drivers, tested on simulators so far.

Scholarly integrity and papers — citation checks, literature/PubMed search, academic experts, PDF tables (6 MCP tools). Panda Federation focused server.

Public leads and enrichment — NPI, podcasts, local businesses, hiring, websites, teams, trends, trucking (8 MCP tools). Panda Federation focused server.

Site, email and agent safety — SPF/DKIM/DMARC, AI/SEO audit, page-change monitor, free hidden-AI scanner (4 MCP tools). Panda Federation focused server.

AzYun is a unified API gateway that works as the model-calling layer behind MCP workflows: one standard API reaches OpenAI, Claude, Gemini and other models, with high-concurrency load balancing and transparent pay-as-you-go billing, so agents and apps can call any model without a separate integration per provider.