Quantum Nexus Discovery API
Discovery, configuration and usage-metrics endpoints.
Discovery, configuration and usage-metrics endpoints.
Every API here is available over the APIs.io API and to AI agents over MCP.
One button, every client — Claude, Cursor, VS Code and the rest.
https://apis.io/mcp
find_apisBrowse and filter every API in the catalog.get_api_artifactsOne API's artifacts, grouped by type.get_openapiThe primary OpenAPI for this API.find_similar_apisAPIs that look like this one.apis_io_searchSTART HERE — APIs, providers and tags for one query, each with its total.resolveTurn a domain, URL or GitHub org into the provider it belongs to.find_cohortsEvery scored population of providers in the catalog.curl "https://apis.io/api/v1/apis/nexusquantum-legal:nexusquantum-legal-discovery-api"
curl "https://apis.io/api/v1/apis?limit=25"
Calls without X-API-Key need no key. Calls with it need a paid plan — get a key.
Free tier, no form to fill in. Signing in shares your email address with us — we store it to create your key and to recognise you if you sign in with another provider. By signing up you are subscribing to the APIs.io and API Evangelist newsletters — you can unsubscribe from any issue. See our Privacy Policy and Terms.
A second provider on the same verified email joins the account you already have.
openapi: 3.2.0
info:
title: Nexus Legal Discovery API
version: 1.95.0
description: 'Multi-jurisdictional legal analysis API with a multi-agent architecture and Zero Retention. ## Output language
`language` (and the language implied by the jurisdiction) sets the language the text is written in. For `es` and `en` the delivered text is also measured: when a substantial part of it came back in the other of the two, a block headed `⚠️ LANGUAGE OF THIS REPORT` (`⚠️ IDIOMA DE ESTE INFORME` when `es` was requested) is added at the end of the text. It works the same with any engine, BYO included. - **When there is no block**: it is not a confirmation. Other languages are not measured, and neither is a text too short or too mixed to tell. - **With BYO**: how closely the output follows the request depends on your engine. - **If the output language is essential to your workflow**: check it on your side against the `analysis` string. ## The verdict travels INSIDE the text — read this before you parse a response
Every string this API returns that a lawyer reads — `analysis`, `answer`, `report`, `draft`, `synthesisText`, and the string fields of structured results like `applicableLaw[]` — carries the citation verdict **inline**, not only in the `citationChecks` object beside it. **What you will actually see:**
```
An acknowledgement does not revive a claim already barred: section 29(7) of the
Limitation Act 1980 【✓】. See also [2019] EWHC 3556 (Comm)
【◻ not checked — the reason is in this citation''s note】. ---
DETERMINISTIC CITATION CHECK (against the official corpus)
1 verified · 1 not checked. ◻ **Not checked** — they were not compared against any official corpus…
```
- **Inline marker**: a `【…】` block immediately after each citation, in the response language. The glyph carries the verdict: `✓` verified · `⚠` a warning (e.g. the decision exists but a party you named does not appear in it) · `⛔` broken · `◻` not checked. - **Footer**: a `---` block at the end with the tally and, for anything not clean, the list of citations and why. **Why it is in the text and not only in the JSON.** Whoever copies an answer into an email, a brief or a .docx takes the PROSE. The `citationChecks` object does not travel with it. A verdict that only exists in a field beside the text is a verdict the reader can lose without noticing. **If you parse these strings**: markers are the only `【…】` sequences we emit, so they are safe to strip with a single regex if your pipeline needs the raw prose — but do not strip them from anything a human will read. `citationChecks` remains the machine-readable source of truth for counts and per-citation status; the markers are the same verdict, placed where it cannot get lost. **Idempotent**: annotating an already-annotated text does not duplicate markers or the footer. ## Streaming: `verifiedText` on the `done` event
On SSE routes the client already holds the raw prose by the time the verifier finishes, so the closing `{type:"done"}` event carries `verifiedText`: the same answer with the verdict already in it. **If it is present, render it in place of the concatenated `text_delta` chunks.** It is omitted when there is nothing to annotate, in which case the deltas are already final. The REST wrappers do this for you — `/v1/consulta`, `/v1/chat` and the rest aggregate server-side and hand you the annotated text — so this only concerns you if you consume the SSE stream directly. ## Base capabilities
- **Mode=deep**: the Node A (Claude Opus 4.7) + Node B (DeepSeek V4-Pro) chain audits adversarially. 2 credits. Returns an `audit` object with structured findings (hallucinations, false citations, omissions). - **Direct models**: calls through the providers'' own APIs (Anthropic, DeepSeek) with no proxies — lower latency, better rate limits. - **Stable Stripe-style error format**: `error.type`, `error.code`, `error.request_id`, `error.details` + backward-compat aliases. ## Errors and rate limits
- **Branch on `error.code`.** A code you do not recognise is handled by its `error.type`.'
contact:
name: Nexus Legal
email: support@nexusquantum.legal
url: https://legal.nexusquantum.legal/developers
license:
name: Commercial
url: https://nexusquantum.legal/terms
servers:
- url: https://legal.nexusquantum.legal
description: Production
security:
- ApiKeyAuth: []
tags:
- name: Discovery
description: Discovery, configuration and usage-metrics endpoints.
paths:
/api/v1/corpus/coverage:
get:
summary: Real-time inventory of the indexed corpus
description: 'Returns the current state of the case-law corpus: number of tables, total documents, jurisdiction inferred per table, date range and top sources with counts. AUTO-DISCOVERABLE: when a new `jurisprudencia_` table with embeddings is added to Supabase, it shows up here with no code changes. Useful for clients (dashboards, MCP, agents) to build coverage UIs without hardcoding anything. Cache: 5 min server-side + `Cache-Control` 5 min on the response. Requires an API key with any valid scope (does not consume credits).'
operationId: getCorpusCoverage
security:
- ApiKeyAuth: []
tags:
- Discovery
responses:
'200':
description: Corpus inventory.
content:
application/json:
schema:
type: object
required:
- generatedAt
- cacheTtlSeconds
- totalDocs
- totalTables
- tables
properties:
generatedAt:
type: string
format: date-time
cacheTtlSeconds:
type: integer
totalDocs:
type: integer
description: Sum of docs across ALL tables.
totalTables:
type: integer
tables:
type: array
items:
type: object
properties:
table:
type: string
example: jurisprudencia_docs
inferredJurisdiction:
type:
- string
- 'null'
example: ES
totalDocs:
type: integer
oldestDoc:
type:
- string
- 'null'
format: date
newestDoc:
type:
- string
- 'null'
format: date
topSources:
type: array
items:
type: object
properties:
source:
type: string
count:
type: integer
sourceColumn:
type:
- string
- 'null'
description: Column used to group topSources.
'401':
description: Invalid API key.
/api/v1/jurisdictions:
get:
summary: List supported jurisdictions
description: Returns the full list of jurisdictions supported by the API, including the legal system of each one. No authentication required.
operationId: listJurisdictions
tags:
- Discovery
responses:
'200':
description: List of jurisdictions.
content:
application/json:
schema:
$ref: '#/components/schemas/JurisdictionsResponse'
/api/v1/usage:
get:
summary: API usage aggregations
description: 'Returns aggregated API usage metrics for the authenticated account. Required by partners that bill their own clients on individual per-sub-key consumption.
Filters: `from`/`to` (ISO date), `granularity` (day/week/month), `api_key_id`, `endpoint`. Format: JSON (default) or CSV with `?format=csv`.'
operationId: getUsage
security:
- ApiKeyAuth: []
tags:
- Discovery
parameters:
- name: from
in: query
schema:
type: string
format: date
description: 'ISO date inclusive (default: 30d ago)'
- name: to
in: query
schema:
type: string
format: date
description: 'ISO date inclusive (default: today)'
- name: granularity
in: query
schema:
type: string
enum:
- day
- week
- month
default: day
- name: api_key_id
in: query
schema:
type: string
format: uuid
description: Filter to a specific sub-key.
- name: endpoint
in: query
schema:
type: string
description: Filter by route (e.g. /api/v1/analyze).
- name: format
in: query
schema:
type: string
enum:
- json
- csv
default: json
responses:
'200':
description: Usage aggregations.
content:
application/json:
schema:
type: object
properties:
period:
type: object
properties:
from:
type: string
format: date
to:
type: string
format: date
granularity:
type: string
enum:
- day
- week
- month
totals:
type: object
properties:
calls:
type: integer
errors:
type: integer
latency_p50_ms:
type: integer
latency_p95_ms:
type: integer
credits_consumed:
type: number
description: Credits consumed (Σ of the `consumption` ledger) in the range, over the key set resolved by ownership.
rate_limited:
type: integer
description: Number of 429 (rate-limited) responses in the range.
by_key:
type: array
description: Breakdown by sub-key.
items:
type: object
properties:
api_key_id:
type:
- string
- 'null'
format: uuid
key_prefix:
type:
- string
- 'null'
calls:
type: integer
errors:
type: integer
credits_consumed:
type: number
description: Credits consumed by this key in the range.
byo_calls:
type: integer
description: Requests served by a BYO engine (routing_policy own_*).
platform_calls:
type: integer
description: Requests served by the platform chain.
rate_limited:
type: integer
description: Number of 429 (rate-limited) responses attributed to this key.
by_endpoint:
type: array
items:
type: object
buckets:
type: array
items:
type: object
text/csv:
schema:
type: string
description: 'CSV dump with columns: period_start,api_key_id,key_prefix,endpoint,calls,errors'
/api/v1/requests/{requestId}:
get:
summary: Read the telemetry of a single request
description: 'Returns the telemetry record (`api_usage_stats`) for a `request_id`. The id is CLIENT-SUPPLIED (echo of the `X-Request-Id` header) and therefore NOT unique across tenants, so the query is ALWAYS scoped to the keys the caller owns:
- **child**: its own `api_key_id` only.
- **master**: the key set of its organization.
- **solo/legacy**: the keys of the caller''s own `user_id`.
The MOST RECENT owned row (`created_at DESC`) is returned as the main object, with `occurrences` = number of owned matches. Uniform 404 (anti-enumeration): a nonexistent `request_id` and one that «exists but is not yours» return exactly the same `NOT_FOUND`, so one tenant cannot probe another''s ids.'
operationId: getRequest
security:
- ApiKeyAuth: []
tags:
- Discovery
parameters:
- name: requestId
in: path
required: true
schema:
type: string
pattern: ^[A-Za-z0-9_-]{8,100}$
description: Request identifier (echo of `X-Request-Id`).
responses:
'200':
description: Telemetry record (most recent owned match).
content:
application/json:
schema:
$ref: '#/components/schemas/RequestRecord'
'401':
description: INVALID_API_KEY.
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'404':
description: NOT_FOUND — no row for that `request_id`, or no row owned by the caller (indistinguishable by design).
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'429':
description: RATE_LIMITED — API key/organization RPM exceeded.
components:
schemas:
JurisdictionsResponse:
type: object
required:
- jurisdictions
- total
- legalSystems
- note
properties:
jurisdictions:
type: array
items:
type: object
description: List of supported jurisdictions with metadata.
total:
type: number
description: Total number of available jurisdictions.
legalSystems:
type: array
items:
type: string
description: Legal systems covered.
note:
type: string
description: 'Where to obtain grounding coverage for the caller''s own entitled jurisdictions. This endpoint deliberately does NOT publish per-jurisdiction corpus state: that is reported by the authenticated `GET /api/v1/normativa/coverage`, scoped to the caller.'
ErrorResponse:
type: object
required:
- error
description: Stripe-style error envelope returned by apiError() on ALL v1 routes. `error` is an OBJECT {type, code, message, request_id, details?}; the top-level `error_message`/`code` are backward-compat (deprecated v1.7). A generated client must deserialize `error` as an object, not as a string.
properties:
error:
type: object
required:
- type
- code
- message
properties:
type:
type: string
example: invalid_request_error
code:
type: string
example: VALIDATION_ERROR
message:
type: string
request_id:
type: string
details:
type: object
description: Optional. An engine failure carries what you need in order to decide what to do next, and a `429` says whose ceiling you reached. Each field is present only when it was measured, and `details` is omitted entirely rather than sent empty.
properties:
retryable:
type: boolean
description: Whether retrying this exact request can succeed. Absent means not measured — do not read absence as `true`.
refunded:
type: boolean
description: Whether the credit for this call was returned. Absent means not measured.
reason:
type: string
description: Short machine-readable cause.
example: provider_timeout
provider:
type: string
description: Which engine failed. On a BYO tenant this is the difference between checking your own key and opening a ticket with us.
example: Anthropic (BYO)
error_class:
type: string
scope:
type: string
enum:
- ip
- user
- organization
description: 'On a `429`: whose ceiling you reached.'
retry_after_seconds:
type: integer
description: 'On a `429`: seconds to wait, the same value as the `Retry-After` header.'
limit_per_minute:
type: integer
description: 'On a `429`: the per-minute ceiling you reached.'
error_message:
type: string
description: Backward-compat (deprecated v1.7).
code:
type: string
description: Backward-compat (deprecated v1.7).
RequestRecord:
type: object
description: Telemetry of one request read from `api_usage_stats`, scoped to the caller's own keys (see `GET /api/v1/requests/{requestId}`).
required:
- request_id
- endpoint
- status_code
- ok
- created_at
- occurrences
properties:
request_id:
type: string
description: Request identifier (echo of `X-Request-Id`).
endpoint:
type: string
example: /api/v1/analyze
status_code:
type: integer
example: 200
ok:
type: boolean
description: status_code < 400.
latency_ms:
type:
- integer
- 'null'
model:
type:
- string
- 'null'
example: DeepSeek V4-Pro+deepseek-v4-pro
description: 'Label of the engine that ANSWERED, not a stable model identifier. It is the chain''s own label — ''DeepSeek V4-Pro'', ''Opus 4.8 · OpenRouter'' — or, when the firm brings its own engine, that engine''s model id. In mode=deep it is composed as ''<analyst>+<auditor>''. `unknown` means nobody could be named: the value is never guessed. Under partner or neutral branding it is replaced by a neutral label. Do NOT key logic on this string: which engine answers is a deployment decision and the chain falls back on its own.'
provider_id:
type:
- string
- 'null'
description: Id of the BYO provider (llm_provider_configs.id) that served the request; null = platform chain.
routing_policy:
type:
- string
- 'null'
enum:
- own_only
- own_then_platform
- platform
- null
description: Effective routing policy.
error_code:
type:
- string
- 'null'
tokens_used:
type:
- integer
- 'null'
input_chars:
type:
- integer
- 'null'
api_key_id:
type:
- string
- 'null'
key_prefix:
type:
- string
- 'null'
example: nlk_ab12
description: Prefix of the matching key (join with `api_keys`).
created_at:
type: string
format: date-time
occurrences:
type: integer
description: Number of your own rows sharing this `request_id`.
requestId:
type: string
description: Id of this endpoint call itself (`X-Request-Id` header).
securitySchemes:
ApiKeyAuth:
type: http
scheme: bearer
bearerFormat: nlk_...