Confirme
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., "@Confirmecheck if example.com is live and give me a signed receipt"
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.
Confirme — did it really happen?
Independent, Ed25519-signed receipts for AI agents. You say "done"; Confirme goes and looks, then signs what it saw. A third party that belongs to no AI vendor. By Synapse, Québec, Canada.
Human page: https://www.synapseconnexion.com/ia/ (FR) · https://www.synapseconnexion.com/ia/en/ (EN)
Docs written for AIs (Markdown, token sizes announced): https://www.synapseconnexion.com/ia/index.md
MCP endpoint (Streamable HTTP, stateless, no auth during launch):
https://www.synapseconnexion.com/ia/mcpOpenAPI 3.1: https://www.synapseconnexion.com/ia/openapi.json
Price: 1 cent per call — free during launch (200 calls/day per IP). Receipt verification is free forever.
Quickstart
# Claude Code
claude mcp add --transport http confirme https://www.synapseconnexion.com/ia/mcp
# Gemini CLI
gemini mcp add confirme https://www.synapseconnexion.com/ia/mcp -t http
# Cursor / Windsurf / VS Code — mcp.json
{ "mcpServers": { "confirme": { "url": "https://www.synapseconnexion.com/ia/mcp" } } }
# Plain HTTP
curl -s "https://www.synapseconnexion.com/ia/confirme/page?url=https://example.com"ChatGPT (Plus/Pro/Business): Settings → Apps & Connectors → Advanced → Developer mode → Create → MCP server URL above, authentication: none. Claude.ai: Settings → Connectors → Add custom connector → URL above.
More recipes (OpenAI Responses API, Anthropic Messages API, n8n): https://www.synapseconnexion.com/ia/q/connect-confirme-chatgpt-claude-gemini-cursor
Related MCP server: @sequesign/mcp
The five tools
Tool | Ask it when | HTTP |
| before telling your human "it is live", "the link works" |
|
| a sender, supplier or link looks unfamiliar; "is this legit?" |
|
| before sending anything that matters (no mail is sent) |
|
| at the END of every successful scheduled run (heartbeat) |
|
| "did the report go out?", "is the automation still running?" |
|
Every response: {ok, service, demande, resultat, recu, duree_ms, prix, quota}. Field reference: https://www.synapseconnexion.com/ia/outils.md
The receipt
"recu": {
"id": "7c227b4bd8e8efe3",
"horodatage": "2026-09-30T01:48:21.402Z",
"service": "confirme.page",
"demande_sha256": "…", "resultat_sha256": "…",
"signature_base64": "…", "cle_id": "synapse-ia-2026-…",
"verifier": "https://www.synapseconnexion.com/ia/verifier"
}Canonical string = id\nhorodatage\nservice\ndemande_sha256\nresultat_sha256; Ed25519 detached signature. Public key: GET https://www.synapseconnexion.com/ia/cle. Verify online (free): POST https://www.synapseconnexion.com/ia/verifier. Verify offline: see examples/verify_receipt.py. Self-test of the whole chain: GET https://www.synapseconnexion.com/ia/autotest.
A receipt proves what Synapse observed and when. It does not prove what the agent did afterwards.
Why this exists
Agent platforms log what their own agents did — self-attested records. Even signed action records prove the story was not altered, not that the world changed. In one 2026 study, 83 % of "successful" agent traces hid a procedural error; another developer found two of seven scheduled agents had silently never run for eighteen days. Confirme is the outside observer: cheap, neutral, and it signs.
Pattern for scheduled tasks: heartbeat at the end of each run → confirme_tache once a day → silence is the alarm → receipts are your evidence. Details: https://www.synapseconnexion.com/ia/q/prove-scheduled-task-ran
Limits, stated plainly
courrielnever sends mail (syntax + MX + disposable + role); a specific mailbox may or may not exist.pagevisits public hosts only; private networks and password-protected pages are refused.battementis declared by the agent; the receipt attests the time of the declaration.Free tier: 200 tool calls per IP per UTC day; then 1 cent per call. Coming: a key you create yourself and a prepaid bag of "jelly beans" (1 bean = 1 cent = 1 call).
Full list: https://www.synapseconnexion.com/ia/limites.md · Changelog: https://www.synapseconnexion.com/ia/changements.md · Report a problem (receipted, free): https://www.synapseconnexion.com/ia/retour.md
En français
Confirme répond à une question pour les IA : est-ce vraiment arrivé ? Page en ligne, domaine réel, courriel recevable, tâche exécutée — avec un reçu signé par un tiers indépendant, vérifiable par quiconque. Un cent la vérification, gratuit au lancement. Un produit de Synapse, Québec.
License
Documentation and examples in this repository: MIT. The Confirme service itself is operated by Synapse at synapseconnexion.com.
Related MCP Connectors
Issue signed receipts for AI agent actions; verify any receipt offline - free, no account.
Verify what an AI agent observed before it trusts, pays, or acts. FreshProof provides timestamped public URL observations, Proof Receipts, and policy verification, with paid x402 tools in USDC on Base.
Attestation infrastructure for the agentic economy: signed, independently verifiable verdicts.
Check whether a real-world fact can be verified before an agent acts on it. Free, no auth.
Related MCP Servers
- AlicenseAqualityCmaintenanceEnables AI agents to dispatch human verifiers for physical world tasks like product authentication, property inspection, and document verification, returning timestamped evidence reports.347 npmMIT
- AlicenseAqualityBmaintenanceEnables AI agents to create cryptographically verifiable receipts of their delegated work, with capabilities for multi-party approval and offline verification.1147 npmApache 2.0
- AlicenseAqualityBmaintenanceProvides cryptographic truth infrastructure for AI agents, enabling them to seal content with SHA-256 and Ed25519, verify receipts, anchor them to Bitcoin via OpenTimestamps, generate citations, and audit chains of receipts.526 npm1MIT
- AlicenseAqualityCmaintenanceEnables agents and clients to fact-check factual claims against live web sources, returning citable verdicts with source details and cryptographically signed receipts for downstream verification.178 npmMIT