Real work history, confirmed by the people on the other side of it, becomes one portable reliability score, owned by the person, not locked in your database, verifiable by anyone. Build on it, add to it, or verify it. One API.
Credda is the portable record of a whole work history, built only from outcomes a third party confirmed, never self-reported claims. A marketplace seller, a professional between jobs, a contractor, a business a buyer relies on, an autonomous agent acting for someone: each is one instance of the same idea, and each is a first-class scored subject. A reputation you earned somewhere should be yours to carry everywhere. The API below is how platforms read that record, add verified outcomes to it, and let the subject present it, and the subject, not the platform, owns it.
One deterministic reliability score per person, with the factor breakdown, six named components, full history and a diff explaining what moved. It belongs to the person, not to your database, so it arrives with them and leaves with them.
See the endpoints →Illustrative: a score rises only as independently verified outcomes accumulate.
People connect GitHub and GitLab to their Credda profile. Your application reads which platforms actually contribute verified evidence, or verifies a signed credential they hand you directly, offline, with no API key at all.
See the endpoints →When an agreement is delivered and a third party confirms it, report that outcome to the append-only ledger. The score recomputes as a pure function of what you reported: idempotent, so retries can never double-count.
See the endpoints →Push outcomes in from the systems you already run, and get HMAC-signed events back the moment a score moves or a band changes. No polling, and every event carries a factor-level reason so your automation can react to why, not just to the number.
See the endpoints →A user mints a revocable share token at a disclosure scope they choose (full, band-only, or minimal) and one script tag renders a live trust badge on any page. Served from this host, resolved against the public verify endpoint, and dead the moment the token is revoked.
See the endpoints →The MCP server gives any Model Context Protocol agent tools to check a counterparty’s trust, verify a credential it was handed offline, or mint and present its own principal’s. One deliberate boundary: AI never decides a score. There is no AI-verdict endpoint, by design.
See the endpoints →Real integrations over the same portable record: a person presenting theirs, a business built by its counterparties, a platform reading a score, an agent checking one. Each is runnable against the live API today, and every step below maps to a documented endpoint.
Let a professional attach verified reliability to a profile or résumé.
Rank a page of sellers by reputation they brought with them.
Read the record a candidate shares with you, without phoning a single reference.
Give a business its own portable record, built by its counterparties.
Let an autonomous agent check who it is dealing with, mid-task.
Official clients, published on npm under the MIT license, each with a public source mirror on GitHub. Webhook signature verification is wire-compatible across all of them.
Typed TypeScript client: score reads, event ingestion, webhooks, and offline credential verification.
npm i @credda/jsLook up, export and verify trust from a terminal or CI. Exit codes are script-friendly.
npm i -g @credda/cliModel Context Protocol server so an AI agent can check a counterparty, or present its own credential, mid-reasoning.
npx -y @credda/mcp-serverThe Go client is published on the Go module proxy: go get github.com/Credda-io/credda-go. The Python client ships in the repo and is not on PyPI yet, so ask us for that one. Every endpoint is a plain HTTPS call, so no SDK is required.
Read a score, then report an outcome. Both endpoints are live right now. Prefer a typed client? npm i @credda/js. All three JavaScript clients (@credda/js, @credda/cli, @credda/mcp-server) are on npm under MIT, and Go is on the module proxy: go get github.com/Credda-io/credda-go. Python ships in the repo, so ask us for that one.
developers# Read a user's reliability score
curl https://api.credda.io/api/v1/users/USER_ID/score \
-H "Authorization: Bearer $CREDDA_API_KEY"
# Report a trust event (idempotent, safe to retry)
curl -X POST https://api.credda.io/api/v1/events \
-H "Authorization: Bearer $CREDDA_API_KEY" \
-H "Idempotency-Key: order-1042-fulfilled" \
-H "Content-Type: application/json" \
-d '{"userId":"USER_ID","eventType":"CONTRACT_FULFILLED","isVerified":true}'Create a key in the console, or browse every endpoint in the API reference.
If you underwrite, release escrow or waive deposits on someone else’s reliability, the whole arc (connect a subject, report outcomes, subscribe to one webhook, read the evidence, decide on your side) is on one page. What a vendor review asks about is answered honestly on trust & compliance.