> ## Documentation Index
> Fetch the complete documentation index at: https://game-narrative.docs.kitefrost.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Set BYOK key

> Store a tenant-supplied LLM provider API key (BYOK) for the project. The key is stored in memory and used instead of the shared platform key for all LLM calls in this project. **Paid plans only** (Pro and above) - see web/public/llms.txt § 'All paid plans include BYOK'. Free-tier tenants in the treatment cohort of the BYOK experiment may also set a key after a successful live probe (KF_FREE_BYOK_COHORT).



## OpenAPI

````yaml /openapi.json put /v1/projects/{project_id}/byok
openapi: 3.1.0
info:
  title: KiteFrost
  description: >
    ## KiteFrost API


    Persistent memory and queryable context for AI-powered products. Create

    projects, track entities and relationships, record events, and query context
    -

    stateful AI you can audit, with durable state across sessions, agents, and

    tools.


    ### What you get


    - **Persistent projects** - entities, relationships, and events live across
    sessions

    - **Context queries** - ask "what does this entity know right now?" and get
    a consistent answer

    - **Event sourcing** - every state transition is recorded and replayable

    - **BYOK** - bring your own LLM provider key per project


    ### Authentication


    All routes (except `/health` and `/ready`) require an `Authorization: Bearer
    <key>` header.

    Use publishable keys (`pk_...`) for read-only client calls and secret keys
    (`sk_...`) for

    write / generation / admin operations.


    ### Async generation pattern


    1. Call a `POST .../generate` endpoint → receive `{"session_id": "...",
    "status": "accepted"}`

    2. Poll `GET /v1/sessions/{session_id}/status` until `status` is `completed`
    or `error`

    3. If `status == "awaiting_human"`, submit a response via `POST
    /v1/sessions/{session_id}/respond`
  version: 1.0.0
servers:
  - url: https://api.kitefrost.ai
    description: Production
security: []
tags:
  - name: projects
    description: >-
      Project, entity, and event management: create projects, upsert entities,
      record events, query context, and generate content with project context.
  - name: npcs
    description: >-
      NPC (non-player character) CRUD, dialogue generation, and export. Supports
      the brainstorm → draft lifecycle.
  - name: sessions
    description: >-
      Poll generation session status, submit HITL responses, and create manual
      state snapshots for rollback/replay.
  - name: events
    description: >-
      Lightweight player-event ingestion path. Accepts player choices, quest
      triggers, and custom events without spinning up a full session.
  - name: players
    description: >-
      Player data management: GDPR erasure (right to be forgotten), data export
      (right of access), player context, facts, quests, and preferences.
  - name: keys
    description: >-
      API key lifecycle management: create publishable / secret keys, list,
      revoke, and rotate keys with scoped permissions.
  - name: byok
    description: >-
      Bring Your Own Key (BYOK) management. Store, revoke, and check the status
      of a tenant-supplied LLM provider API key per project.
  - name: ttrpg
    description: >-
      TTRPG GM pack endpoints: encounter CRUD and generation, session CRUD and
      async generation.
  - name: chapters
    description: >-
      Book Author pack endpoints: chapter CRUD, async generation, and markdown
      export.
  - name: ideas
    description: >-
      Idea Generation Engine: generate novel story ideas, list, fetch, mutate,
      blend, assign, and archive ideas.
  - name: seeds
    description: >-
      Story-seed discovery operations: list, detail, on-demand generate, promote
      (canonical assign per R127-C6), lifecycle transition, and soft-delete.
  - name: feedback
    description: >-
      Collect explicit feedback signals (thumbs, rating, A/B choice) and
      retrieve aggregated insights for model improvement.
  - name: webhooks
    description: >-
      Webhook subscription management: register HTTPS endpoints for async event
      delivery, list, delete, and test HMAC signatures.
  - name: audio
    description: >-
      Audio Story pack endpoints: async audio-story generation, list, fetch, and
      export (SSML / manifest / audio).
  - name: billing
    description: >-
      Billing integration: checkout, portal, subscription management, tier info,
      and usage stats.
  - name: demo
    description: >-
      Public demo endpoint - try the KiteFrost from the landing page without
      signing up. Limited to 5 messages per session and 20 requests per IP per
      hour. No authentication required.
  - name: health
    description: >-
      Health and readiness probes for load-balancer and orchestrator liveness
      checks.
  - name: score
    description: >-
      Content scoring: run craft-layer quality metrics (C10-C15) and humanness
      scoring (H2) on arbitrary text to get a per-dimension quality breakdown.
  - name: resilience
    description: >-
      Resilience Score: aggregate compliance, training, and education pack
      results into an overall project resilience score (0-100) with
      per-dimension breakdown and trend.
  - name: compliance-exercises
    description: >-
      Compliance Tabletop pack endpoints: create exercises, generate scenarios,
      evaluate participant responses, and produce after-action reviews.
  - name: training
    description: >-
      Training Simulation pack endpoints: create scenarios, generate branching
      scenario trees, and assess learner competency.
  - name: support
    description: >-
      Support Orchestration pack endpoints: create ticket contexts, generate
      response drafts, and escalate tickets.
  - name: story_bible
    description: >-
      Story Bible: assemble the Knowledge Graph into a structured canonical
      reference of characters, locations, events, factions, and items with their
      relationships. Supports point-in-time snapshots via at_seq.
  - name: gamification
    description: >-
      Gamification endpoints: player points, earned achievements, leaderboards
      with configurable visibility and weekly reset, and activity streaks.
  - name: conversations
    description: >-
      Multi-player conversation lifecycle: create, join, submit input, stream
      events (SSE), leave, and get conversation state.
  - name: images
    description: >-
      Image generation: create AI-generated images from text prompts with
      pack-configurable presets, safety controls, and audit logging.
paths:
  /v1/projects/{project_id}/byok:
    put:
      tags:
        - byok
      summary: Set BYOK key
      description: >-
        Store a tenant-supplied LLM provider API key (BYOK) for the project. The
        key is stored in memory and used instead of the shared platform key for
        all LLM calls in this project. **Paid plans only** (Pro and above) - see
        web/public/llms.txt § 'All paid plans include BYOK'. Free-tier tenants
        in the treatment cohort of the BYOK experiment may also set a key after
        a successful live probe (KF_FREE_BYOK_COHORT).
      operationId: set_byok_key_v1_projects__project_id__byok_put
      requestBody:
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/SetBYOKRequest'
        required: true
      responses:
        '200':
          description: BYOK key stored
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/BYOKStatusResponse'
        '401':
          description: Invalid or missing API key
        '402':
          description: BYOK requires a paid plan - upgrade to Pro or higher
        '403':
          description: Forbidden - API key lacks the required scope.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/APIError'
              example:
                error: Forbidden
                code: forbidden
                request_id: 3fa85f64-5717-4562-b3fc-2c963f66afa6
                message: Your API key does not have the required scope.
                doc_url: https://docs.kitefrost.ai/errors/forbidden
        '404':
          description: Not Found - resource does not exist or is not accessible.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/APIError'
              example:
                error: Not found
                code: not_found
                request_id: 3fa85f64-5717-4562-b3fc-2c963f66afa6
                message: The requested resource does not exist.
                doc_url: https://docs.kitefrost.ai/errors/not_found
        '422':
          description: Validation Error
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/HTTPValidationError'
        '429':
          description: Rate limit exceeded
        '500':
          description: >-
            Internal Server Error - unexpected server-side failure. Retry with
            exponential back-off.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/APIError'
              example:
                error: Internal server error
                code: internal_error
                request_id: 3fa85f64-5717-4562-b3fc-2c963f66afa6
                message: An unexpected error occurred.
                doc_url: https://docs.kitefrost.ai/errors/internal_error
components:
  schemas:
    SetBYOKRequest:
      properties:
        api_key:
          type: string
          minLength: 1
          title: Api Key
          description: LLM provider API key
        provider:
          type: string
          title: Provider
          description: LLM provider name (e.g. openai, anthropic)
          default: openai
        provider_url:
          anyOf:
            - type: string
            - type: 'null'
          title: Provider Url
          description: >-
            Optional custom LLM provider base URL (e.g. for self-hosted
            OpenAI-compatible endpoints). Must be an HTTPS URL that passes SSRF
            validation - internal network addresses, cloud metadata endpoints,
            and non-HTTPS schemes are rejected.
        model_preference:
          anyOf:
            - type: string
            - type: 'null'
          title: Model Preference
          description: >-
            Optional specific model the tenant prefers from their provider (a
            provider model id). Used as a hint when routing - if the model
            cannot satisfy the request's capability requirements, normal model
            selection is applied.
        allow_platform_fallback:
          type: boolean
          title: Allow Platform Fallback
          description: >-
            When ``False``, requests that cannot be satisfied by the tenant's
            BYOK key fail with an explicit error instead of silently falling
            back to the platform's shared key. Defaults to ``True`` (permissive)
            to preserve current behavior.
          default: true
      type: object
      required:
        - api_key
      title: SetBYOKRequest
      description: PUT /v1/projects/{project_id}/byok
    BYOKStatusResponse:
      properties:
        project_id:
          type: string
          title: Project Id
        has_key:
          type: boolean
          title: Has Key
        source:
          type: string
          title: Source
        provider:
          anyOf:
            - type: string
            - type: 'null'
          title: Provider
        key_preview:
          anyOf:
            - type: string
            - type: 'null'
          title: Key Preview
        model_preference:
          anyOf:
            - type: string
            - type: 'null'
          title: Model Preference
        allow_platform_fallback:
          type: boolean
          title: Allow Platform Fallback
          default: true
        probe_status:
          anyOf:
            - type: string
            - type: 'null'
          title: Probe Status
        model_warning:
          anyOf:
            - type: string
            - type: 'null'
          title: Model Warning
      type: object
      required:
        - project_id
        - has_key
        - source
      title: BYOKStatusResponse
      description: BYOK key status (never exposes the actual key).
    APIError:
      properties:
        error:
          type: string
          title: Error
        code:
          type: string
          title: Code
        request_id:
          anyOf:
            - type: string
            - type: 'null'
          title: Request Id
        detail:
          anyOf:
            - additionalProperties: true
              type: object
            - type: 'null'
          title: Detail
        message:
          anyOf:
            - type: string
            - type: 'null'
          title: Message
        doc_url:
          anyOf:
            - type: string
            - type: 'null'
          title: Doc Url
        feedback_id:
          anyOf:
            - type: string
            - type: 'null'
          title: Feedback Id
      type: object
      required:
        - error
        - code
      title: APIError
      example:
        code: not_found
        doc_url: https://docs.kitefrost.ai/errors/not_found
        error: Not found
        message: >-
          The requested resource does not exist or you do not have permission to
          access it. Verify the resource ID and your API key scopes.
        request_id: 3fa85f64-5717-4562-b3fc-2c963f66afa6
    HTTPValidationError:
      properties:
        detail:
          items:
            $ref: '#/components/schemas/ValidationError'
          type: array
          title: Detail
      type: object
      title: HTTPValidationError
    ValidationError:
      properties:
        loc:
          items:
            anyOf:
              - type: string
              - type: integer
          type: array
          title: Location
        msg:
          type: string
          title: Message
        type:
          type: string
          title: Error Type
        input:
          title: Input
        ctx:
          type: object
          title: Context
      type: object
      required:
        - loc
        - msg
        - type
      title: ValidationError

````