---
title: "Get Guideline"
method: GET
path: "/marketing/guidelines/documents/{file_id}"
tags: ["marketing-guidelines"]
---

# Get Guideline

`GET /marketing/guidelines/documents/{file_id}`

One guideline's requirements, in the order they were extracted.

Polled by the detail view while a decomposition runs, which is exactly when the previous
generation has just been retired — so `state` is not a convenience. Every regeneration leaves
another permanent generation behind, and without a filter here the payload carries all of them and
the browser discards most of it, on every poll. `superseded_with_human_edits` is counted separately
so the surface can still offer to show them.

Carries no presigned URL: signing is not idempotent, so a URL on a polled payload changes every
tick and remounts the document viewer. The viewer gets its URL from `usePreviewUrl`.

## Path parameters

- `file_id` string, uuid, required

## Query parameters

- `state` GuidelineRequirementState[], nullable

## Response `200`

Successful Response

- GuidelineDetailResponse
  - `guideline` GuidelineSummary, required — A marketing standards document and the state of its requirement library.
    - `file_id` string, uuid, required
    - `filename` string, nullable, required
    - `upload_id` string, uuid, nullable, required
    - `newer_version_processing` boolean, required
    - `active_requirement_count` integer, required
    - `total_requirement_count` integer, required
    - `proposed_requirement_count` integer
    - `needs_requirements` boolean, required
    - `stale` boolean, required
    - `latest_decomposition` DecompositionSummary, required — One extraction run, and how it went. `status` is the *effective* status: a run whose worker never reported back reads as failed rather than as permanently running, so the surface never offers a spinner that will not resolve.
      - `decomposition_id` string, uuid, required
      - `upload_id` string, uuid, required
      - `status` 'queued' | 'running' | 'succeeded' | 'failed', required — Lifecycle of one decomposition run.
      - `stage` 'queued' | 'downloading' | 'extracting' | 'grounding' | 'storing' | 'done', required — Where a running decomposition has got to. Only the stages a watching human can perceive. Resolving the model's channel names to `MarketingCollateralType` and writing the rows are both sub-second, so they are not stages — a step that never renders is a step that misleads. `EXTRACTING` is the one that takes the minute.
      - `queue_wait_seconds` integer, nullable, required
      - `requirements_total` integer, nullable, required
      - `requirements_written` integer, required
      - `page_count` integer, nullable, required
      - `error` string, nullable, required
      - `started_at` string, date-time, nullable, required
      - `finished_at` string, date-time, nullable, required
      - `requested_by_email` string, nullable, required
      - `superseded` boolean, required
  - `requirements` GuidelineRequirementResponse[], required
    - `requirement_id` string, uuid, required
    - `file_id` string, uuid, required
    - `filename` string, nullable, required
    - `decomposition_id` string, uuid, nullable, required
    - `guideline_set_id` string, uuid, nullable, required
    - `ordinal` integer, required
    - `revision_number` integer, required
    - `state` 'active' | 'disabled' | 'archived' | 'superseded' | 'proposed', required — Whether a requirement is in the working library, and if not, why not. All of them are recorded as revisions, so every one is reversible and readable. ARCHIVED and SUPERSEDED are deliberately distinct: ARCHIVED is a person deciding this obligation does not apply to them, SUPERSEDED is a newer decomposition of the same guideline taking over. A regeneration must never overwrite the first with the second — that would erase a human decision.
    - `requirement_text` string, required
    - `judging_guidance` string, nullable, required
    - `context` string, nullable, required
    - `collateral_types` MarketingCollateralType[], nullable, required
    - `marketing_team_ids` string[], nullable, required
    - `source_collateral_terms` string[], nullable, required
    - `trigger_terms` string[], required
    - `unresolved_collateral_terms` string[], required
    - `section_label` string, nullable, required
    - `title` string, nullable, required
    - `source_page` integer, nullable, required
    - `citation` GroundedSpan, required — Where a requirement appears to have come from, and how much to trust it.
      - `page_number` integer, required
      - `boxes` BoundingBox[]
        - `x0` number, required
        - `y0` number, required
        - `x1` number, required
        - `y1` number, required
      - `source_text` string, nullable
      - `matched_terms` string[]
      - `score` number
      - `basis` string
      - `cited_page` integer, nullable
      - `refusal` string, nullable
    - `change_kind` 'extracted' | 'edited' | 'disabled' | 'enabled' | 'archived' | 'restored' | 'superseded' | 'proposed' | 'approved', required — What a revision was for. Redundant with comparing a revision to its predecessor, and that is the point: the reason a change was made is the part a diff cannot recover.
    - `change_note` string, nullable, required
    - `updated_at` string, date-time, required
    - `updated_by_email` string, nullable, required
    - `has_human_edits` boolean, required
    - `signal` RequirementSignal — One requirement's track record, as a ranked read reports it. Present only when the list was asked for an `order`; `None` otherwise, so an ordinary library read pays neither the aggregate nor the payload. A row whose requirement has never been checked is absent from a ranked read entirely rather than carrying zeros — there is nothing to rank. **The two halves are in different units and must not be added or compared.** `collateral_checked` and the three `checks_*` count *checks* — one per (evaluation, requirement), which `uq_marketing_requirement_check_once_per_evaluation` guarantees. The `risks_*` count *risks*, and one check can produce several, so `risks_flagged + risks_over_flagged` is not bounded by `checks_failed`. The rates come from the server because the server ordered by them. A client re-deriving them would be a second definition of the number the ranking was built on. `None` means no evidence, which is a different claim from `0.0`.
      - `collateral_checked` integer, required
      - `checks_failed` integer, required
      - `checks_passed` integer, required
      - `checks_not_determined` integer, required
      - `checks_current_revision` integer, required
      - `verdicts` object, required
      - `risks_superseded` integer, required
      - `false_positive_rate` number, nullable, required
      - `dismissal_rate` number, nullable, required
      - `fire_rate` number, nullable, required
    - `downvotes` integer, nullable
  - `superseded_with_human_edits` integer, required

## Other responses

- `422` — Validation Error

## Changes

- **2026-09-02** `425d5b8a3c17` — 1 breaking, 3 warning, 4 info
  - response property `requirements/items/decomposition_id` list-of-types was widened by adding types `null` to media type `application/json` of response `200`
  - added the new `approved` enum value to the `requirements/items/change_kind` response property for the response status `200`
  - added the new `proposed` enum value to the `requirements/items/change_kind` response property for the response status `200`
  - added the new `proposed` enum value to the `requirements/items/state` response property for the response status `200`
  - …4 more

[Change history](https://skmtc.dev/kobaltlabs/apis/fastapi/changes/marketing/guidelines/documents/:file_id/get.md)

---

[API](https://skmtc.dev/kobaltlabs/apis/fastapi.md) · [All operations](https://skmtc.dev/kobaltlabs/apis/fastapi/llms.txt) · [OpenAPI document](https://skmtc-service-production.skmtc.workers.dev/v1/apis/kobaltlabs/fastapi/revisions/425d5b8a3c17/schema)