Skip to main content
Glama
tools-that-dont-upload

tools-that-dont-upload

Tools That Don't Upload

Listed on mcpservers.org

MCP server and REST API for data, text, code, PDF and web tasks. Requests are processed in memory and not stored.

Korean tools

Romanization, amounts in Korean words (일금 ...원정), particles (조사), initial consonants (초성), business-number checksum, personal-data masking. Document generation: 견적서 / 거래명세서 (XLSX), 차용증 / 표준근로계약서 (DOCX).

Related MCP server: llm-gateway

Egress receipts

Every REST response has an X-Egress header and every MCP result has _meta["com.toolsthatdontupload/egress"] listing external hosts contacted (none for most tools).

Files through MCP

PDF tools accept file_base64 / files_base64 (max 10 MB) and return file_base64.

Quick start

Claude Code:

claude mcp add --transport http tools-that-dont-upload https://toolsthatdontupload.com/mcp

Cursor / other clients:

{"mcpServers":{"tools-that-dont-upload":{"url":"https://toolsthatdontupload.com/mcp"}}}

Limits

  • Anonymous: compute tools under fair use; file 20 calls/day; AI 10 calls/day per IP.

  • Free key: 1,000 credits/month. Paid plans: coming soon.

Privacy

Yellow = edge memory only. Orange = our edge fetches the URL you give. Blue = Cloudflare Workers AI. Details: https://toolsthatdontupload.com/privacy/

Note: this repository is a documentation snapshot, not the full deployed source.

Listed in the Claude Market MCP directory

MIT License

Repository layout

  • workers/ — source of the production Cloudflare Workers that this repo deploys: tools-mcp-server (the MCP server at /mcp), tools-hub-ratelimit (gateway for network, file and AI tools), tools-hub-keys, tools-hub-kr (Korean tools), tools-hub-docs (Korean documents). Every folder in workers/ is deployed by GitHub Actions (.github/workflows/deploy.yml → scripts/deploy.mjs) with build provenance attestations; the live commit is shown at https://toolsthatdontupload.com/mcp and https://toolsthatdontupload.com/transparency/.

  • tests/ — black-box regression tests that run against a deployed MCP endpoint (node tests/search-regression.mjs [MCP_URL]).

  • plugin/ — Claude plugin (hosted connector + offline local server + 2 skills).

  • local/ — offline MCP bundle source (.mcpb, release local-v1.0.0).

  • server.json — MCP Registry metadata.

  • SECURITY.md — how outbound requests are restricted, known limits, and how to report issues.

Development notes

  • The MCP server code that actually runs is workers/tools-mcp-server/worker.mjs. The old root mcp-server.js was an outdated copy and was removed on 2026-10-12 (still in git history).

  • Not everything is in this repo: the static site, the ~25 compute-tool Workers and the network-tool Workers are deployed separately. The MCP server reaches them through Cloudflare service bindings.

  • Pushing to workers/, scripts/deploy.mjs or the workflow file deploys to production. Test changes on a separate workers.dev script first.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables AI assistants to parse and search documents including PDF, Word, Excel, PowerPoint, and images via OCR, with support for semantic search and batch processing.
      2
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables users to run shell commands with compressed output, extract information from large files, drive local opencode and Codex agents, perform browser-automation tasks, and monitor runs via a live dashboard.
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables agents to search, resolve, and request authorized integration actions on demand while keeping the full Pipedream catalog server-side, so teams can act on time-aware meeting memory and shared context without context overload.
      MIT
    • F
      license
      Not graded
      quality
      A
      maintenance
      Enables vector search and document retrieval, extraction and chunking of files into Markdown, and private deep research generation from a configured pipeline. Users can query knowledge bases, convert documents, and produce reports through natural language.
      -