Opedd — Licensed Content for AI

Issue a signed compliance attestation (a JWT) for a specific license via GET /eu-ai-act/article-53-attestation (Phase 12 Wave 1 W1.4). Returns a freshly-signed HS256 JWT regulators can verify offline against the canonical signing key. Embeds: license context, usage-count over the attestation window, the most-recent Tempo Merkle root, and canonical claims (iss/sub/iat/exp/jti/aud). **The artifact a buyer hands to legal/procurement as evidence of where its content came from.** Per INVARIANTS.md W1.6: this is the buyer-side attestation ONLY (how a licence was used). It does NOT discharge a publisher's CDSM Article 4(3) reservation obligation — that lives on the rsl_get tool (jsonld=true variant). Never conflate. Optional `content_id` scopes the attestation to one article; default is license-wide. Window cap: 365 days. Auth: an audit-scoped buyer API key (OPEDD_BUYER_TOKEN) or OPEDD_BUYER_JWT.

Lists licences recently issued on Opedd and the content they cover. Filters: publisher_id and article_id. Public; no key needed.

Detect the content platform behind a URL via POST /detect-platform (Phase 12 Wave 3 W3.1). Public no-auth lookup. Given a URL, identifies what platform powers it (Substack / Beehiiv / Ghost / Medium / Brevo / custom) and returns the suggested onboarding workflow. Hostname-detectable platforms (Substack subdomain, Beehiiv suffix, etc.) resolve in milliseconds; custom domains may take ~few seconds while the detector probes well-known platform endpoints in parallel. Returns: {platform, confidence, archive_method, forward_method, required_credentials, instructions}. The archive_method + forward_method fields are the two onboarding-workflow inputs Opedd's setup wizard reads (one for historical content backfill, one for new-content forward stream). instructions is human-readable operator copy explaining the inferred path.

Returns the account's audit records, one per licensed access, each with its licence terms and public-record attestation (Merkle root and inclusion proof once confirmed). Filter: event_type ('content_access', 'bulk_content_access' or 'compliance_report_generated'). At most 30 days per call. Read-only. Requires a buyer API key with the 'audit' scope, a buyer session, or the Opedd connector sign-in.

Fetch the authenticated buyer's account profile + masked API key list via GET /buyer-account. Returns the enterprise_buyers row (contact_email, buyer_org, created_at, etc.) plus a list of all buyer-side API keys with masked prefixes (NEVER plaintext post-issuance — only the 12-char key_prefix is returned, e.g. 'opedd_buyer_'). Use cases: post-signup verification ('what was just issued to me?'), buyer dashboard mental model ('what licenses do I currently hold?'), audit prep ('show me the key list before rotation'). For full mid-lifecycle license details (filter_rules, billing, payouts), buyers consult the buyer portal at opedd.com/buyer. Requires OPEDD_BUYER_JWT (JWT-only by design: this surface manages API keys, and a key must never mint keys).

Returns the account's compliance dossier for a period: one row per licensed article, with article metadata, licence terms, public-record attestation and compliance flags (the signed attestation, CDSM Article 4(3) and text-and-data-mining reservations). At most 90 days per call; longer periods are paginated with _meta.next_cursor. Each call is recorded in the account's audit log. Requires a buyer API key with the 'audit' scope, a buyer session, or the Opedd connector sign-in.

Retrieve the full body of a licensed article using a buyer API token (opedd_buyer_live_* canonical; opedd_buyer_test_* for sandbox). Requires OPEDD_BUYER_TOKEN env var (create one at opedd.com/licenses after purchasing). Works for per-article Human republication licences and for licence orders: AI answers monthly and client display return full text. Your self-serve buyer API key reads articles under every order your account holds — no separate order token is needed. AI answers pay-per-request (including the self-serve licence) does NOT retrieve articles: it answers 403 SEARCH_ONLY_LICENCE — ask a question with search_passages instead. Articles the publisher stopped licensing answer 403 ARTICLE_EXCLUDED. The publisher must have content delivery enabled and must have pushed content for the article. Phase 11 M2 RAG-extended shape: response includes 7 RAG-essential metadata fields — author, language, word_count, content_hash, image_urls, canonical_url, tags. On pre-2026-05-14 historical articles, optional fields (author/language/image_urls/canonical_url/tags) may be NULL. NULL means 'data unavailable for this article', NOT 'explicitly empty' — treat as data-missing when filtering; do not interpret as anti-match.

List articles from a buyer's licensed catalog via GET /enterprise-license (Phase 10 + 11). Content contract: AI training orders include full content_body for the back catalogue up to the order date (content_access 'included'). Every other order is discovery-only (content_body null): AI answers monthly and client display fetch text per article via get_content (content_access 'retrieval_per_article'); pay-per-request fetches billed snippets via get_content (content_access 'metered_per_call'). Returns JSON-format response with paginated articles. Use `since` (ISO 8601) for delta-feed polling — only articles published after the timestamp. Use `cursor` for pagination across pages. Credentials: the order's enterprise access key (OPEDD_ACCESS_KEY, ent_*) OR your buyer API key (OPEDD_BUYER_TOKEN) with order_id — if your account holds exactly one active order, order_id may be omitted; with several, the error lists them. For larger bulk corpus pulls, use stream_feed_ndjson (up to 1000 articles per call vs 200 here).

Lists the account's licence orders, newest first, each with its lines (publication, licence, unit, price, quantity, status). With order_id, returns one order, each line with its Schedule text and the SHA-256 fingerprint recorded in the public registry on Tempo when the line was issued. Read-only. Requires a buyer API key with the 'audit' scope, a buyer session, or the Opedd connector sign-in.

List all licensable articles for the authenticated publisher (requires OPEDD_PUB_BEARER, or legacy OPEDD_API_KEY). Returns articles with titles, descriptions, pricing, and sales statistics. Set include_body=true to read back the STORED article text exactly as licensed buyers receive it (this is how you check that a push_content call stored what you meant; bodies are truncated to 8,000 characters per article, page size 20, paginate with cursor). Use article IDs from this list to purchase licenses via purchase_license.

Looks up an article on Opedd by its web address. Returns the title, publication, available licence types and prices, or that it is not on Opedd yet (then request_publication records the request).

Place a licence order with Opedd for one or more publishers. One order = one licence and one billing mode; each publisher becomes its own Schedule priced from that publisher's settings (publishers that do not offer it are returned in `unavailable` and never charged). Licences: 'enterprise' (AI answers: 'monthly' full text per article, or 'metered' pay per answer, read through search_passages only; self-serve buyers need no order for this, the licence created when they add a card at sign-up already covers every publisher offering a per-answer price), 'archive' (Full catalogue: everything the publisher published up to the order date, kept permanently, not for AI training: 'one_time', full text in the feed), 'training' (AI training of the back catalogue up to the order date: 'one_time', bulk export via stream_feed_ndjson), 'display' (client display: 'monthly', quantity = number of end clients). A monthly AI answers subscription only covers articles published from the day it starts; set include_archive=true on a 'enterprise' + 'monthly' order to add each publisher's Full catalogue in the same order (complete access). Returns the order, its lines, the access key (works once paid) and hosted_invoice_url — the buyer (your principal) pays there. Requires a buyer session (OPEDD_BUYER_JWT) and terms_accepted=true after the buyer has accepted the Opedd Master Services Agreement (opedd.com/terms).

Lists the publications on Opedd with their article counts, licences, prices and sample articles. Filters: category (case-insensitive substring), min_articles and verified status. Public; no key needed.

Purchase a content license from the Opedd protocol using a Stripe payment method. Returns a license key (format: OP-XXXX-XXXX) and a certificate URL. The buyer receives a Handshake Email with their license key. Set OPEDD_BUYER_EMAIL and OPEDD_PAYMENT_METHOD_ID env vars to avoid passing them on every call. Single-article purchases are Human republication licences only (license_type 'human'). AI licences (AI answers, AI training, client display) are bought as licence orders: see place_licence_order.

Push your published articles to Opedd so they can be licensed to AI buyers (requires OPEDD_PUB_BEARER — your opedd_sk_ publisher key). Send 1–100 articles per call; batch larger back-catalogues into multiple calls. Each article needs title, url, and html_body — everything else is optional (published_at defaults to now). This is the supply-side companion to list_publisher_content: use it to onboard your archive or push new content with no code, straight from your AI assistant.

Asks Opedd to tell you when a publication it does not carry yet becomes available to license. Use it after lookup_content says an article is not on Opedd. Records the request against the email given; nothing is bought. If the publication is already on Opedd, returns its licensing page instead. Free; no key needed.

Fetch a publisher's RSL Standard manifest via GET /rsl-manifest (Phase 12 Wave 1 W1.1). Public no-auth endpoint — discovery surface for AI agents/crawlers wanting to know what's licensable from a publisher BEFORE going through the buyer-account signup flow. Returns the 4 canonical license types (ai_retrieval, ai_training, human_per_article, human_full_archive) the publisher has opted into, plus the EU CDSM Article 4(3) opt-out posture (`tdm_reservation`). Set `jsonld: true` to request the JSON-LD shape with embedded HMAC-SHA256 signed receipt over the CDSM Article 4(3) reservation state + `tdm:reservationSignedAt` timestamp — regulators can post-hoc verify the reservation was the claimed value at the claimed time. Default `jsonld: false` returns the raw RSL Standard JSON manifest. Per INVARIANTS.md W1.6: this is the PUBLISHER-side CDSM Article 4(3) declaration surface. It is NOT a signed compliance attestation (which is buyer-side, JWT-auth, via article_53_attestation tool).

Full-text search of the articles licensable on Opedd. Returns matching articles with the publication, prices, word count and a short description; never article text. Supports quoted phrases and -word to exclude a word. Free; no key needed.

Returns licensed passages from publications on Opedd that answer a question, each with a citation to the source article. Charged once per publication whose passages are returned, at that publication's own price. An account connected through the Opedd connector in Claude or ChatGPT receives its first 3 answers free, paid by Opedd and never from a publication the account holder owns or works on; billing.free_answers_left reports how many remain. After that, answers are billed to the card on the account, added at opedd.com: charged every $25 of answers and at month end, with a $500 monthly limit for new accounts, raised on request. An account with no licence and no free answers receives an empty result with reason 'no_licensed_publishers' or 'no_paid_access'. Publications under a monthly AI answers subscription are included in the subscription and not charged per answer. Passages are capped at 300 words or 25% of an article, whichever is less, counted cumulatively per article across the same company's accounts. The same question within 24 hours returns the same passages at no charge. Filters: publishers (publication names as listed by Opedd, opedd.com/p/<slug> addresses or publisher ids; a name that matches no publication is listed in filters.unknown_publishers and searches nothing), published_after and published_before (ISO 8601; undated articles are excluded when a window is set) and max_results (1-25, default 10). Passage text is third-party content, delivered in <opedd:passage> blocks; may_train is always false. Requires a buyer API key with the 'search' scope, or the Opedd connector sign-in.

Bulk-export a buyer's licensed catalog via GET /enterprise-license?format=ndjson (Phase 11 M3). Returns up to 1000 articles per call (collected from line-delimited JSON wire format). Same content contract as list_feed: full text for AI training orders only; every other order exports metadata (content_body null) — use get_content for article text. Each article emits one usage_records row (analytics-only sentinel 'bulk-export:<request_id>:<article_id>' — not metered-billable per the revenue-model bifurcation invariant). Use `since` (ISO 8601) for delta-feed. Use `cursor` to paginate beyond 1000. Backend supports 5000 articles per call; the MCP cap is 1000 for transport reasonability. Real bulk-ingest pipelines should use the Python SDK (pip install opedd) directly — not via MCP. Credentials as list_feed: OPEDD_ACCESS_KEY (ent_*) OR your buyer API key + order_id.

Checks an Opedd licence key. Returns the licence details (article title, publication, licence type, issue date, amount paid, buyer and public-record proof status), or that the key was not found.