An AI agent can verify a student's transcript, courses, and witness co-signatures with no login — over the Model Context Protocol or a single plain HTTPS call. We believe it's the first record built to be verified by an agent, not just read by one, for K-12, homeschool, and alternative pathways.
Admissions offices and HR teams are wiring AI into their pipelines — application readers that ingest and summarize uploaded PDFs. But a PDF is an unverified claim. The agent extracts text and hopes it's true.
Eformogi is the record an agent can actually trust: hash-anchored to a public ledger, witness co-signed, and verifiable at a URL. Instead of guessing whether a transcript is real, the agent checks it — and reads who stood behind it.
Two ways to verify the same record. Both are live on production right now.
A GET against the public verify endpoint returns non-identifying proof metadata for any issued receipt:
# Public. No key, no login. (Free for single-link verifies.) curl "https://eformogi.com/api/verify-receipt?id=tr_8e59bb9641e8"
The tr_8e59bb9641e8 id above is a real, live ledger receipt for a clearly-fictional demo student — run the curl and you'll get this response back. To see the receiver view in your browser, open /verify/tr_8e59bb9641e8 (no real student — the record says so itself).
Response shape:
{
"verified": true,
"kind": "family_transcript",
"id": "tr_8e59bb9641e8",
"issued_at": "2026-07-21T01:36:44Z",
"hash": "8e59bb9641e87b7d080a99a2dd99595b…",
"student_name": "Jordan Sample (Fictional)",
"issuer": "Sample Family Records Office — fictional demo record",
"program": "Homeschool · mixed-provider — family-graded, mastery, co-op, dual-enrollment & AP (Demo)",
"grad_year": "2027",
"course_count": 12,
"total_credits": 10,
"tier_counts": { "self": 3, "family": 4, "mentor": 4, "receiver": 1 },
"course_signature_rollup": { "total": 12, "co_signed": 4 },
"courses": [ { "cr_receipt_id": "cr_0365c42ab299", … } ],
"issued_by": "Eformogi"
}For agents that already own an Eformogi key, the MCP server exposes six policy-governed tools over JSON-RPC 2.0. Here's a read_profile call:
curl -X POST https://eformogi.com/api/mcp \
-H "Authorization: Bearer efg_YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "read_profile",
"arguments": { "categories": ["education"] }
}
}'Full setup (Claude Desktop config, policy model, all six tools) lives in the MCP docs. The REST contract for /api/verify-receipt and the rest of the agent surface is in the developer docs.
read_profileRead vault data for specified categories, filtered by the owner's per-category policy.list_consentsList the owner's active consent receipts — who they have shared with.create_consentCreate a consent grant. Categories under "ask" queue for the owner's approval.check_policyReport which categories the agent may access (auto / ask / never).log_actionLog an explicit action description into the owner's activity feed.search_orgsSearch organizations the owner can share data with.A parent builds a transcript in Eformogi and issues it. The record is hashed (SHA-256) and written to a public ledger. It gets an id like tr_8e59bb9641e8.
A coach, evaluator, or teacher counter-signs individual courses or work samples. Each signature is its own verifiable receipt. Disagreements are shown, never hidden.
An admissions AI, an HR agent, or a person pastes the id — over MCP or a plain HTTPS GET. They get back the hash, the issuer, the course count, and the co-signature rollup. No Eformogi account required.
Machine-readable entry points, all live:
Claude is a trademark of Anthropic, PBC. ChatGPT is a trademark of OpenAI, Inc. Gemini is a trademark of Google LLC. Model Context Protocol is an open standard published by Anthropic. Eformogi is an independent product and is not affiliated with, endorsed by, or sponsored by Anthropic, OpenAI, or Google.