Reach Gateway
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Reach Gatewaywhat's the health of the evidence channels?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Reach Gateway
Reach Gateway is an owner-authenticated, read-only service for normalized public evidence. It runs on Cloudflare Workers, uses Cloudflare Access OAuth, and exposes the registered private app Reach the G to ChatGPT and Codex through MCP and a Custom GPT Action facade.
Discovery stays with the host assistant. Reach uses no paid search API and provides no private-account or write access.
Actions
Action | Purpose |
| Check Web, X, YouTube, Reddit, and RSS evidence channels. |
| Retrieve one public HTTPS URL as inert, provenance-bearing evidence. |
| Retrieve available public captions for a supported YouTube URL. |
Production MCP endpoint: https://reach-gateway.izeesub.workers.dev/mcp
ChatGPT's header-constrained private adapter uses
https://reach-gateway.izeesub.workers.dev/mcp-token/<owner-secret>. The path
token is validated by the Worker against REACH_GATEWAY_TOKEN; the transport
may be configured as No Auth in ChatGPT without making the server public.
Cloudflare Access OAuth remains available in parallel on /mcp.
Custom GPT schema: https://reach-gateway.izeesub.workers.dev/openapi.json
The Action bundle and exact GPT Builder instructions are under actions. Its operation IDs are checkEvidenceChannels, readPublicUrl, and getPublicTranscript; all three call the same retrieval core as MCP.
Related MCP server: Genius Sports RSS Monitor MCP
Local verification
Requires Node.js 24 or newer.
npm ci
npm run check
node scripts/verify-no-secrets.mjs
npx wrangler deploy --dry-runCredentialed source tests are separate and do not certify a release when skipped:
REACH_LIVE_TESTS=1 npm run test:liveSecurity boundary
Exact-owner OAuth is enforced by Cloudflare Access and Worker-side identity verification.
Retrieval accepts only public HTTPS destinations and passes through SSRF-safe URL policy.
Retrieved content is labeled and treated as untrusted evidence.
Cookies, bearer tokens, private-account credentials, writes, and paid search keys are outside the product contract.
The service is stateless and does not create a research-memory authority.
Private distribution
The Codex bundle is under plugin/reach-gateway and maps to the existing registered app instead of creating a duplicate MCP identity. The release is distributed through the owner's personal plugin catalog. Public multi-user directory submission is intentionally unavailable because reusable reviewer access would conflict with owner-only OTP/MFA.
See docs/evidence/release-verification.md for exact verification state and remaining device checks.
Rollback
Use one explicit, previously verified Cloudflare deployment ID:
./scripts/rollback.sh <deployment-id>Rollback is a production mutation and must remain human-authorized.
This server cannot be deployed
Maintenance
Related MCP Connectors
Read-only Remote MCP for externally grounded AI agent trust receipts.
Verified business OSS MCP for search, RSS, crawling, documents, browser, media and transcription.
Judged, citation-checked policy corpus over MCP. Keyless public reads; API key for AI tools.
Read-only hosted MCP over CanonicAI's cited Answers corpus on canonicai.com.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceA read-only MCP server that gives AI agents the web as compact, ranked, verified evidence — no API keys, no cloud retrieval, all models local.13 npm1MIT
- FlicenseNot gradedqualityCmaintenanceRead-only MCP server for monitoring public RSS and Atom feeds from competitor and industry websites, enabling competitive intelligence workflows through feed fetching, searching, and thematic summarization.-
- AlicenseCqualityCmaintenanceMCP server for safely reading public URLs for AI agents, providing tools to fetch, extract, cache, and inspect web content as evidence.15MIT
- FlicenseAqualityBmaintenancePersistent knowledge-base MCP server. Semantic search with source citations over saved YouTube videos, podcasts, articles and PDFs — one endpoint for every MCP-capable AI client.28-