generated: '2026-09-07'
method: searched
source: https://docs.getelva.ai/quality-and-scoring/score-a-spec-via-api
authentication:
  style: none (public scoring endpoints); OAuth 2.0 + PKCE for the platform MCP resource
  reference: authentication/elva-authentication.yml
content_types:
  request:
    - text/plain (raw OpenAPI YAML or JSON, streamed as the body)
    - application/json (parsed spec object)
  response: application/json; charset=utf-8
versioning:
  style: none-visible
  note: >-
    Public paths are unversioned (/api/review, /api/review/checks). No versioning or
    deprecation policy is published for the public API.
error_envelope:
  shape: '{"error": "<message>"}'
  statuses: [400, 422]
  reference: errors/elva-problem-types.yml
idempotency:
  coverage: na
  note: >-
    The public API has no state-mutating surface: POST /api/review is a stateless
    scoring computation (submit a spec, receive grades) with nothing persisted and
    no resource created, and GET /api/review/checks is a read. No Idempotency-Key
    mechanism is documented, and none is needed — repeating either call is naturally
    safe. na, not none: there is no write surface for a mechanism to cover.
reversibility:
  coverage: na
  note: >-
    No write surface — no operation creates, mutates, or deletes provider-side state,
    so there is nothing to reverse. Scoring calls can simply be re-run.
dry_run_mode:
  coverage: na
  note: No write surface; the scoring call is itself consequence-free.
pagination:
  style: none
  note: GET /api/review/checks returns the full flat array in one response.
request_tracing:
  note: No request-id header documented; live responses carry an ETag but no trace id.
rate_limit_signaling:
  headers: []
  note: >-
    No X-RateLimit-*/RateLimit-*/Retry-After headers observed on live responses
    (probed 2026-09-07) and none documented for the public scoring API.
  reference: rate-limits/elva-rate-limits.yml
scoring_semantics:
  note: >-
    Each check reports Pass/Partial/Fail scored 1.0/0.5/0.0 times its weight; category
    percentages are weighted pass ratios; overall score is the category-weighted average
    (Design 40, DeveloperExperience 35, AIReadiness 15, Security 10). Grades A >= 88,
    B >= 76, C >= 64, D >= 50, else F.