> ## Documentation Index > Fetch the complete documentation index at: https://docs.alakazam.gg/llms.txt > Use this file to discover all available pages before exploring further. # Edit the graph from natural language > Run the kernel agent server-side: describe the change in prose and the agent authors it, then the SAME fail-closed kernel gate every deterministic write passes must bless the result before it persists. This is real LLM spend — one unit is reserved against your daily `edit` quota BEFORE the agent runs, and refunded on a validation reject, agent error, or concurrency conflict. RESULT-ONLY: the response is exactly `world`, `diagnostics`, and `reply` — the agent's system prompt, op grammar, lexicon, model id, and raw op trace are never exposed. If the agent needs clarification it returns the question in `reply` and leaves the world unchanged. Pass an `Idempotency-Key` so a retry doesn't double-charge. ## OpenAPI ````yaml /alakazam-v1.yaml post /v1/worlds/{id}/edit openapi: 3.1.0 info: title: Alakazam — Programmable Worlds API version: '1.0' description: > **Alakazam is the programmable worlds API.** Generate playable, AI-rendered worlds from a prompt or an image, program their logic, and embed them in your own products and games. A world is *programmable*: a live graph of states and events you generate, then read, edit, fork, drive, and react to through the API. Two surfaces: a **Creation API** (generate, read, manage, fork worlds) and a **Runtime/Embed API** (mint short-lived session tokens your end-users' browsers use to boot an embedded world). **Authentication.** Two schemes. Management endpoints (`/v1/apps*`) use your Supabase user session. Data endpoints use an **API key** issued per app: `pk_…` (publishable, browser-safe, read + embed) and `sk_…` (secret, server-only, create worlds + mint sessions). Never ship a secret key to a browser. **Usage & quota.** Generations and session mints are metered per app and reserved before any GPU spend; exceeding the daily quota returns `402`. servers: - url: https://api.alakazam.gg description: Production (placeholder — set to your conjure-service host) security: [] tags: - name: Apps & Keys description: >- Create apps (tenants) and manage API keys. Authenticated with your user session. - name: Worlds description: >- Create, read, manage, and fork SMWorld games. Authenticated with an API key. - name: Graph editing description: > Read and program a world's graph — states (nodes), events (edges), and the entrance — with deterministic CRUD, a batch op vocabulary, a natural-language kernel-agent edit, and the kernel validate/lint gate. Every write is validated fail-closed before it persists. Authenticated with an API key. - name: Versions description: > Snapshot, branch, check out, and diff a world's graph. Versions form a branching tree of full snapshots; a HEAD pointer tracks the working graph. Authenticated with an API key. - name: Characters description: > Create, manage, and talk to characters — a SMWorld subtype that pairs a stance-graph with a "brain" (persona, lore, voice). CRUD is authenticated with an API key; the live talk turn (/say) and voice (/tts) are called from the browser with a short-lived session token. - name: Sessions description: Mint short-lived runtime tokens for embedding a world. - name: Usage description: Per-app usage and quota. - name: Webhooks description: >- Register HTTPS endpoints to receive signed server-side event notifications. Managed with your user session. paths: /v1/worlds/{id}/edit: post: tags: - Graph editing summary: Edit the graph from natural language description: > Run the kernel agent server-side: describe the change in prose and the agent authors it, then the SAME fail-closed kernel gate every deterministic write passes must bless the result before it persists. This is real LLM spend — one unit is reserved against your daily `edit` quota BEFORE the agent runs, and refunded on a validation reject, agent error, or concurrency conflict. RESULT-ONLY: the response is exactly `world`, `diagnostics`, and `reply` — the agent's system prompt, op grammar, lexicon, model id, and raw op trace are never exposed. If the agent needs clarification it returns the question in `reply` and leaves the world unchanged. Pass an `Idempotency-Key` so a retry doesn't double-charge. parameters: - name: id in: path required: true schema: type: string format: uuid - name: If-Match in: header required: false schema: type: string description: The `rev` you last read. Mismatch returns 409. - name: Idempotency-Key in: header required: false schema: type: string description: >- A retried edit with the same key returns the original result without a second LLM spend. requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/EditRequest' responses: '200': description: >- Edit applied (or a clarifying question in `reply` with no change). The new `rev` is the `ETag`. content: application/json: schema: $ref: '#/components/schemas/EditResult' '400': description: Missing instruction content: application/json: schema: $ref: '#/components/schemas/Error' '401': description: Authentication required content: application/json: schema: $ref: '#/components/schemas/Error' '402': description: Daily edit quota exceeded content: application/json: schema: $ref: '#/components/schemas/Error' '404': description: World not found content: application/json: schema: $ref: '#/components/schemas/Error' '409': description: Concurrent modification — reload and retry (reservation refunded) content: application/json: schema: $ref: '#/components/schemas/Error' '422': description: >- The agent's world failed the kernel gate (not persisted; reservation refunded) content: application/json: schema: $ref: '#/components/schemas/GraphValidationError' '502': description: Authoring agent failed (reservation refunded) content: application/json: schema: $ref: '#/components/schemas/Error' '503': description: Authoring or quota backend unavailable (fail-closed) content: application/json: schema: $ref: '#/components/schemas/Error' security: - ApiKeyAuth: [] components: schemas: EditRequest: type: object description: > A natural-language edit instruction for the server-side kernel agent. Either `instruction` or `message` is required. properties: instruction: type: string description: >- What to change, in prose — e.g. "add a locked vault behind the cellar and wire a key to open it". message: type: string description: Alias for `instruction` (accepted for convenience). verifiedObjects: type: array description: Optional grounding hints (named objects the agent may reference). items: type: object additionalProperties: true expectedRev: type: string description: >- Optional optimistic-concurrency token; alternative to the `If-Match` header. EditResult: type: object description: > The RESULT-ONLY response from a kernel-agent edit. It carries the validated `world`, advisory `diagnostics`, and the agent's natural-language `reply`, and NOTHING else — the agent's system prompt, op grammar, lexicon, model id, and raw op trace are never exposed. When the agent needs clarification it returns its question in `reply` and leaves the world unchanged. required: - world - diagnostics - reply properties: world: $ref: '#/components/schemas/SMWorld' diagnostics: type: array description: >- The full validateWorld set (advisory warning/info findings) for the persisted world. items: $ref: '#/components/schemas/KernelDiagnostic' reply: type: string description: >- The agent's natural-language summary of what it did, or a clarifying question when no change was made. Error: type: object properties: detail: type: string description: Human-readable error message. errors: type: array items: type: string description: Field-level validation errors (e.g. on 422 from POST /v1/worlds). schemaVersion: type: string required: - detail GraphValidationError: type: object description: >- A 422 from a graph write or validate — the world failed the kernel gate. Nothing was persisted. required: - detail - diagnostics properties: detail: type: string diagnostics: type: array items: $ref: '#/components/schemas/KernelDiagnostic' schemaVersion: type: string example: '1.0' SMWorld: type: object description: > A playable world. The full contract is published as JSON Schema at https://alakazam.gg/schema/smworld-1.0.json (see schema/COMPATIBILITY.md). Additional optional properties are allowed (forward-compatible). additionalProperties: true required: - id - name - entrance - scene properties: id: type: string name: type: string description: type: string cover: type: string entrance: type: object required: - image - state properties: image: type: object required: - src properties: label: type: string src: type: string description: Seed image for the boot. state: type: string description: Initial state id. scene: type: object required: - states - events properties: states: type: object additionalProperties: true events: type: array items: type: object additionalProperties: true KernelDiagnostic: type: object description: > A single finding from the kernel grammar + doctrine lint suite. STRUCTURAL lints (e.g. `dangling-ref`, `transition-needs-to`, `kernel-cycle`, `slot-*`) are always fatal `error`s and block the write (422). Doctrine/budget lints (e.g. `negation`, `whiteout`, `lethal-override`, `budget`) are advisory and surface as `warning`/`info`. required: - lint - severity - path - message properties: lint: type: string description: The lint id (e.g. `dangling-ref`, `negation`, `budget`). severity: type: string enum: - error - warning - info path: type: string description: Human-readable location, e.g. `event[Step outside].movement.static`. message: type: string description: What is wrong and, where possible, how to fix it. loc: type: object description: Structured location for editor/narrator UIs. properties: field: type: string template: type: string slot: type: string fill: type: string securitySchemes: ApiKeyAuth: type: http scheme: bearer description: >- An app API key, e.g. `Authorization: Bearer sk_live_…`. Secret (`sk_`) for writes/sessions; publishable (`pk_`) for read/embed. ````