Agent Tools MCP
RemoteAbout Agent Tools MCP
Agent Tools MCP is a remote collection of utilities for data validation, text optimization and synthetic test-data generation. It offers a complete endpoint and three focused connectors for MCP clients.
Key features
- 01Validate common data formats and values
- 02Minify HTML, JavaScript and CSS
- 03Optimize datasets, tags and prompts
- 04Generate synthetic data and identifiers
- 05Use grouped validators, optimizers or generators
Use cases
- 01Check user-entered data formats
- 02Prepare text and assets for processing
- 03Create fixtures for development tests
MCP support
The remote MCP server exposes 32 tools over Streamable HTTP. Clients can use the full endpoint or connect to specialized validator, optimizer and generator endpoints.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://agent-tools-mcp.yvan11.workers.dev/mcpclaude mcp add --transport http why-see-agent-tools https://agent-tools-mcp.yvan11.workers.dev/mcp{"mcpServers":{"why-see-agent-tools":{"type":"http","url":"https://agent-tools-mcp.yvan11.workers.dev/mcp"}}}Open in Cursor{"servers":{"why-see-agent-tools":{"type":"http","url":"https://agent-tools-mcp.yvan11.workers.dev/mcp"}}}Open in VS Code{"mcpServers":{"why-see-agent-tools":{"type":"http","url":"https://agent-tools-mcp.yvan11.workers.dev/mcp"}}}{"mcpServers":{"why-see-agent-tools":{"serverUrl":"https://agent-tools-mcp.yvan11.workers.dev/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEAgent Tools MCP
32 free remote MCP tools for validation, optimization and synthetic test data.
Website · 32-tool directory · Health · Registry metadata · MIT License
Connect an agent
Streamable HTTP (all 32 tools): https://agent-tools-mcp.yvan11.workers.dev/mcp
Specialized connectors (same Cloudflare Worker; no extra deploys or credentials):
| Connector | Tools | URL |
|---|---|---|
| Validators MCP | 12 | https://agent-tools-mcp.yvan11.workers.dev/mcp/validators |
| Optimizers MCP | 7 | https://agent-tools-mcp.yvan11.workers.dev/mcp/optimizers |
| Generators MCP | 13 | https://agent-tools-mcp.yvan11.workers.dev/mcp/generators |
Choose the smallest relevant connector. All endpoints share the same rate limits, privacy cautions, source and deployments.
Claude Code:
claude mcp add --transport http agent-tools https://agent-tools-mcp.yvan11.workers.dev/mcp
VS Code / Copilot: Command Palette → MCP: Add Server → HTTP → paste the endpoint URL.
Claude Code project .mcp.json configuration:
{ "mcpServers": { "agent-tools": { "type": "http", "url": "https://agent-tools-mcp.yvan11.workers.dev/mcp" } } }
VS Code .vscode/mcp.json configuration:
{ "servers": { "agent-tools": { "type": "http", "url": "https://agent-tools-mcp.yvan11.workers.dev/mcp" } } }
Use an MCP client rather than a browser GET request to /mcp, which can return Method not allowed by design.
Security and publishing
This is a public endpoint without authentication. Never submit real passwords, payment-card data, tax identifiers, bank information, private documents, or API secrets. Synthetic generated IDs and seeded passwords are development fixtures only. Read SECURITY.md.
The repo is public and MIT licensed, preserving the vendored source licenses. Cloudflare Workers Builds tests and deploys on commits to main; no GitHub Actions. The official registry descriptor uses io.github.why-see/agent-tools-mcp; Glama metadata identifies the maintainer. See PUBLISHING.md for a no-local-repository submission procedure. These files do not automatically publish a registry listing.
Tool inventory
Validators (12): is-card-safe, is-email-safe, is-iban-safe, is-iso-date-safe, is-ip-safe, is-json-safe, is-password-safe, is-phone-e164, is-uuid-safe, is-url-safe, is-us-tin-safe, is-vat-safe.
Optimizers (7 tools / 6 modules): optimize-dataset, optimize-google-tag, minify-html, minify-js, minify-css, minify-prompt, strip-ansi.
Generators (13): generate-avatar, generate-card-number, generate-color, generate-dataset, generate-fake-data, generate-hash, generate-lorem, generate-password, generate-phone-number, generate-prompt, generate-slug, generate-ssn, generate-token.
Optimizer MCP tool names are action-oriented, while vendored upstream file paths retain their original names. Implementations are derived from the upstream module files. Exact upstream versions and update policy are documented in UPSTREAM.md. The upstream catalogs are not complete or consistent enough to serve as the runtime source of truth; we use explicit audited function adapters.
Tests and deployment
On GitHub, push to main → Cloudflare Workers Builds performs npm test, then npx wrangler deploy only if tests pass. No GitHub Actions are present.
The test suite invokes all 32 MCP tools through the actual handler, including the initialize, tools/list, and tools/call protocol methods. It also verifies the three specialized MCP endpoints, their isolated inventories, and backward compatibility with /mcp.
The Worker has a Cloudflare native rate-limit binding (30 requests per minute per IP per data center), 8-KiB body guard and per-tool input/output bounds.
Optional MCP_API_TOKEN must be a Cloudflare secret, not a GitHub secret or a variable in wrangler.toml.
Restrictions
This public endpoint runs pure software utilities and cannot access your GitHub, Cloudflare, OpenAI or payment accounts. It is not a remotely editable filesystem. Do not send real passwords, payment details or private personal data to its validators. Generated financial identifiers and SSNs are synthetic test fixtures only. This service does not provide OAuth, client billing, SLA, global quotas or user identity.
Current catalog size is 32 MCP tool names backed by 31 original upstream modules. An update to an upstream source repo does not automatically rebuild this MCP Worker; see the explicit update policy.
The original four endpoint URLs are unchanged. The seven optimizer operation names were standardized to action-oriented forms in October 2026; clients that explicitly call a previous is-* optimizer name must refresh their tool catalog. The underlying code, inputs, outputs and tool counts are unchanged.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.