# QED Proof > Independent verification of AI agents' work: signed receipts, an append-only log, anchored on Base. - [QED Proof](https://docs.qedproof.site/index.md): Independent verification of AI agents' work — signed receipts, an append-only log, anchored on Base. - [Quickstart](https://docs.qedproof.site/quickstart.md): Connect GitHub, create an API key, and send your first claim to QED Proof in a few minutes. - [Claims](https://docs.qedproof.site/concepts/claims.md): What an agent submits to QED Proof — always untrusted input, checked against the destination. - [Verdicts](https://docs.qedproof.site/concepts/verdicts.md): The five outcomes QED Proof can reach when it checks a claim against its destination. - [Receipts](https://docs.qedproof.site/concepts/receipts.md): The signed, tamper-evident record of a claim, what QED Proof observed, and the resulting verdict. - [Merkle log](https://docs.qedproof.site/concepts/merkle-log.md): The append-only, RFC 6962-style log every receipt is written into. - [Anchoring](https://docs.qedproof.site/concepts/anchoring.md): How the Merkle log's root is committed on a public chain, so it can't be backdated or silently rewritten. - [Trust levels](https://docs.qedproof.site/concepts/trust-levels.md): What a checker can actually confirm about a receipt, from a signature alone up to a hardware- or web-proven observation. - [Check a receipt yourself](https://docs.qedproof.site/concepts/check-a-receipt.md): Verify a receipt's signature, Merkle inclusion, and on-chain anchor independently, without trusting QED Proof's API. - [Connectors](https://docs.qedproof.site/connectors/overview.md): How QED Proof connects to a destination, what each connector reads, and which ones can verify a claim today. - [GitHub](https://docs.qedproof.site/connectors/github.md): Connect GitHub read-only and verify that a commit, a pull request or a CI check really landed. - [http.url.status](https://docs.qedproof.site/connectors/http.md): Check that a public URL returns the status code — and optionally the exact content — your agent claimed, no connector required. - [Facebook Pages, Instagram, Meta Ads](https://docs.qedproof.site/connectors/meta.md): Connect Facebook Pages, Instagram and Meta Ads read-only today. Verification for these connectors isn't built yet. - [Slack](https://docs.qedproof.site/connectors/slack.md): Connect Slack read-only today. Verification of messages your agents post there isn't built yet. - [X](https://docs.qedproof.site/connectors/x.md): Connect X read-only today. Verification of posts your agents publish there isn't built yet. - [Coming soon](https://docs.qedproof.site/connectors/coming-soon.md): WhatsApp and Gmail aren't connectable yet. - [Clients](https://docs.qedproof.site/clients.md): What exists today for sending claims and checking receipts, and what's planned. - [Alerts](https://docs.qedproof.site/alerts.md): Send verdicts to a Discord webhook as they're decided. - [Pipelines](https://docs.qedproof.site/pipelines.md): A planned way to define your own proof around the actions that matter to you, not just the actions QED Proof already checks. - [Self-hosting](https://docs.qedproof.site/self-hosting/overview.md): Run your own POAW node — the same code the hosted product runs on. - [Docker Compose](https://docs.qedproof.site/self-hosting/docker-compose.md): Run PostgreSQL and the node with one command. - [The poaw-node CLI](https://docs.qedproof.site/self-hosting/cli.md): Run a POAW node without Docker — schema, keys, workspaces and the server, driven by poaw-node and environment variables. - [Security](https://docs.qedproof.site/security.md): Data minimization, read-only connections, and how to report a vulnerability. - [Introduction](https://docs.qedproof.site/api-reference/introduction.md): The QED Proof API — submit claims, read claim status and receipts, and fetch the issuer's public keys. - [Authentication](https://docs.qedproof.site/api-reference/authentication.md): How QED Proof API keys work, and which endpoints need one. - [Errors](https://docs.qedproof.site/api-reference/errors.md): How the QED Proof API reports failures — an HTTP status code plus a plain-text detail message. - [Reason codes](https://docs.qedproof.site/api-reference/reason-codes.md): The reason_code values that accompany a receipt's verdict, and which verdicts they go with. - [Submit a claim](https://docs.qedproof.site/api-reference/submit-a-claim.md): Tells QED Proof what an agent says it did. The claim is checked against the destination itself, inline when the destination answers quickly, otherwise by the next worker pass. Submitting the same `client_claim_id` again returns the existing claim (`created: false`) instead of a second one. - [Get a claim](https://docs.qedproof.site/api-reference/get-a-claim.md): The claim's current status and, once decided, its verdict and receipt id. - [Get a receipt](https://docs.qedproof.site/api-reference/get-a-receipt.md): A signed receipt with its inclusion proof in the log (and the anchor, once the log root is anchored). Receipts are meant to be shared and checked by anyone, so this needs no API key. Reads are rate-limited per IP address. - [Get the public keys](https://docs.qedproof.site/api-reference/get-the-public-keys.md): The receipt-signing public keys, and the anchor attesters and schemas a checker must accept (SPEC §5.2, §8.4). Public by design. ## OpenAPI Specs - [openapi](/docs/api-reference/openapi.json)