Entity Bind / Resolvd API quickstart — offer planned Read without an account: /api/docs Static navigation and request access /api/pricing Subscription plans, limits and planned evaluation offer /api/openapi.json Generated shipped schema and revision fingerprints /api/mcp Complete MCP connection and tool guide /api/mcp/tools.json Generated MCP tools/list schemas and metadata /api/operations Auth, errors, rate limits, retries and long-running work /api/sandbox Executable isolated local sandbox with a dated offline feed /api/openai-agents Runnable OpenAI integration and investigation workspace /docs and /redoc Interactive Swagger and browsable API reference /api/demo Finite snapshots, never arbitrary live execution /api/evidence Methodology and evidence limitations /llms.txt Agent index /llms-full.txt Full corpus with current generated OpenAPI Request evaluation access: mailto:support@tenhaw.com?subject=Entity%20Bind%20evaluation%20access The owner authorizes an agent and receives a resolve-scoped key via the issued-key flow. No instant key is promised while launch gates are blocked. Use the actual API origin and MCP URL supplied by your deployment; do not guess hosted origins. With an authorized key, set RESOLVD_API_ORIGIN and RESOLVD_API_KEY in your local secret environment. First make a free, authenticated readiness read: curl --fail-with-body "$RESOLVD_API_ORIGIN/v1/reference/enrichment-fields" \ -H "Authorization: Bearer $RESOLVD_API_KEY" This resolution request performs computation and can use credits: curl --fail-with-body "$RESOLVD_API_ORIGIN/v1/resolve" \ -H "Authorization: Bearer $RESOLVD_API_KEY" \ -H 'Content-Type: application/json' \ -H 'Idempotency-Key: own-evaluation-smoke-1' \ --data '{"name":"INSTITUTE OF PATENTEES & INVENTORS","country":"GB", "registration_number":"00170887","jurisdiction":"GB","depth":"research"}' Read /v1/usage, /v1/reference/limits and /v1/reference/enrichment-fields using that key for actual balances/source availability/rights. Read each response's /v1/decisions/{decision_id} for its authorized receipt. Stateless /v1/match has no entity ID or resolve receipt; repeated calls are fresh work unless they are retries of the same durable operation. Same idempotency key with changed content is a conflict. Different resolve company content is fresh computation. MCP uses the actual deployment MCP origin and the same resolve-scoped bearer key. resolve_company supports the company context and structured address schema; REST and MCP share the tenant grant. The same key cannot administer accounts or approve reviewer decisions. Runtime source readiness remains authenticated. Use /api/mcp for copyable connection examples and current tool schemas; /api/operations explains typed errors, Retry-After, idempotency and long tasks. This deployment's shared production batch limit: 500 records. Batch admission must include row/internal-comparison exposure, not one standalone 5-credit ceiling. Evaluate Entity Bind on your own companies. We are preparing 1,000 one-time credits with full research depth, no card and 30 days from your first admitted resolution-family computation. Request evaluation access; availability requires confirmation. Status: planned; instant self-service activation is not live. Planned grant: 1,000 lifetime credits and 1,000 lifetime deep attempts per eligible verified organization/account, shared by its tenant's keys, REST and MCP. No renewal, automatic purchase, card charge or overage. The separate Free plan remains 300 credits with probabilistic maximum depth. Chargeable shallow outcomes use 1 credit; adjudication/research outcomes use 5. Abstentions, no identifying signal and eligible exact-content resolve replays use 0. Chargeable new_entity outcomes retain their existing billing. Credits are not response counts: up to 1,000 shallow or 200 deep chargeable outcomes, or a mixture. Deep attempts include abstentions. Reads consume no credits. New research-capable operations reserve 5 credits; with only 1-4 available, explicitly choose depth=probabilistic (1-credit ceiling) or receive an evaluation_insufficient_credits refusal. Never silently lower depth. Deep-attempt exhaustion, expiry and provider-spend pauses are distinct refusals. Confirmed eligible free replay and authorized evidence reads survive expiry, subject to published retention and erasure. Upgrade needs explicit purchase. Proposed evaluation throughput: 60 resolve/match operations per minute and two concurrent deep executions per tenant. Shared production source/request/batch limits also apply. Monetary ceilings and expected pause behavior require an operator decision and measured pilot costs before activation. No numerical provider-spend ceiling has been approved by this repository implementation. Evidence methodology New resolve, compare and batch submissions require a nonblank company `name` and valid `country` (country name or ISO code), even when a legal identifier is supplied. Optional `description` maps to `context`; `postcode` maps to `address.postal_code`. Conflicting alias and canonical values are rejected. Existing-catalogue reads retain optional filters. Freeze user-chosen cases and independently expected authority-qualified company numbers or LEIs before execution. Include exact identifiers, legal-name variants, name/domain, postcode, country, typed city, town prose, business descriptions, common-name collisions, brands/subsidiaries, stale or misleading context and true no-matches. Keep unknown truth unknown. Test cold and explicitly seeded catalogues separately; a namespace reset must never replenish evaluation entitlement. Inspect decision, match_confidence, match_confidence_basis, entity_created, clarification, shortlist and receipt together. A newly minted entity ID alone proves no real-world identity. Candidate-backed accepted matches require >=0.90; confidence is population-calibrated evidence, not independent accuracy proof. Check each returned field's source, time, currency/period, company/group scope and licence. Missing fields remain absent; no commercial data rights are bundled. Export every outcome, including errors, ambiguities and abstentions. Report coverage, independently verified precision, unknown truth, false binds, latency, field fill/freshness and credit/provider cost separately with denominators and company-family grouping. Compare identical frozen inputs and source access/date. Never discard difficult cases after seeing results or prescribe a recommendation. The internal regression accepts 499/705. The 99% minimum-input target remains unmet. Historical UK replay's 1,666/1,740 includes legacy/missing calls and no full-prompt-verified calls. Neither these results nor this small fixture supports an independent current customer holdout or a global 'best' claim. The starter runner is scripts/run_own_case_eval.py in the repository. Its freeze command accepts CSV/JSON. Its run command consumes the immutable manifest and exports all cases; offline records require matching request hashes. Live execution requires the account owner's issued key and an explicit --allow-live option. Do not impersonate a person to obtain access. A headless agent can use an issued resolve-scoped bearer key; MCP connection and OAuth authorization are documented at /api/mcp. Public researching agents need no identity for these docs/examples. Data retention: evidence_retained_until is a minimum of 30 days after evaluation expiry, subject to authorized erasure/revocation. Automatic maintenance purge skips all evaluation tenants to preserve evidence through expiry extensions. There is no automatic evaluation-data deletion schedule; the owner can explicitly erase data. A dynamic deletion policy and minimal eligibility/abuse retention remain launch decisions. No scheduled deletion promise is made here.