> ## Documentation Index > Fetch the complete documentation index at: https://docs.seneca.professionalaiagents.com/llms.txt > Use this file to discover all available pages before exploring further. # Create research > Accept a research question. A new run returns 202 and reserves the panel price. The same Idempotency-Key with the same prompt and panel returns 200 for 24 hours, sets Idempotent-Replayed: true, and does not reserve credits again. ## OpenAPI ````yaml https://seneca-research.fly.dev/v1/openapi.json post /v1/research openapi: 3.1.0 info: title: Seneca Research API version: 0.1.0 description: >- Create a research run, poll it, then read the customer report. Original model text is a separate read: list /v1/research/{run_id}/responses, read one model at /v1/research/{run_id}/responses/text, or download the panel as JSONL from /v1/research/{run_id}/responses/export. Send Authorization: Bearer seneca_sk_… Narrative quickstart: https://seneca.professionalaiagents.com/api.md servers: - url: https://seneca-research.fly.dev description: Production security: [] paths: /v1/research: post: tags: - public-api summary: Create research description: >- Accept a research question. A new run returns 202 and reserves the panel price. The same Idempotency-Key with the same prompt and panel returns 200 for 24 hours, sets Idempotent-Replayed: true, and does not reserve credits again. operationId: create_research_v1_research_post parameters: - name: Idempotency-Key in: header required: true schema: type: string minLength: 1 maxLength: 255 pattern: ^[A-Za-z0-9._:-]{1,255}$ description: >- Required. 1–255 characters: letters, numbers, or . _ : -. Reuse this exact value, with the same prompt and panel, to retry a request whose response was lost. A different prompt or panel with this value returns 409 idempotency_conflict. The key expires after 24 hours. description: >- Required. 1–255 characters: letters, numbers, or . _ : -. Reuse this exact value, with the same prompt and panel, to retry a request whose response was lost. A different prompt or panel with this value returns 409 idempotency_conflict. The key expires after 24 hours. requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/PublicResearchCreate' responses: '200': description: >- Idempotent replay of the original run. The response header Idempotent-Replayed is true. No second run is created and no second reserve is written. headers: Idempotent-Replayed: description: true when this response replays an existing run. schema: type: string const: 'true' content: application/json: schema: $ref: '#/components/schemas/PublicRunOut' '202': description: >- Research accepted. Store id and poll GET /v1/research/{id} until status is completed, partial, or failed. content: application/json: schema: $ref: '#/components/schemas/PublicRunOut' '400': description: >- invalid_request, idempotency_key_required, invalid_idempotency_key, or quick_pulse_unavailable. A malformed body is invalid_request and does not reserve credits. quick_pulse_unavailable rejects a new Quick Pulse launch and does not reserve credits. content: application/json: schema: $ref: '#/components/schemas/PublicErrorResponse' '401': description: >- authentication_required when the bearer token is missing or not a bearer token. invalid_api_key when the key is unknown, revoked, or expired. content: application/json: schema: $ref: '#/components/schemas/PublicErrorResponse' '402': description: >- insufficient_credits or subscription_required. insufficient_credits includes details.required and details.available. content: application/json: schema: $ref: '#/components/schemas/PublicErrorResponse' '409': description: >- idempotency_conflict when Idempotency-Key is reused with a different prompt or panel. concurrency_limit_reached when the workspace already has the maximum number of runs in progress. content: application/json: schema: $ref: '#/components/schemas/PublicErrorResponse' '429': description: >- rate_limit_exceeded. Creates are limited to 30 requests per minute per workspace. content: application/json: schema: $ref: '#/components/schemas/PublicErrorResponse' '500': description: internal_error. The body does not include a traceback or a secret. content: application/json: schema: $ref: '#/components/schemas/PublicErrorResponse' security: - SenecaApiKey: [] components: schemas: PublicResearchCreate: properties: prompt: type: string title: Prompt description: The research question. The server trims surrounding whitespace. panel: type: string title: Panel description: >- Research depth. Canonical ids and server-owned prices: wide (15 credits, 10 curated models), consensus (45 credits, 40 curated models), seneca (100 credits, every production-eligible model). The server chooses the models. Extra fields, including a model list, are rejected. additionalProperties: false type: object required: - prompt - panel title: PublicResearchCreate PublicRunOut: properties: id: type: string format: uuid title: Id object: type: string const: research.run title: Object default: research.run status: type: string title: Status description: queued, running, synthesizing, completed, partial, or failed. phase: anyOf: - type: string - type: 'null' title: Phase description: >- panel while models are still answering. Null while synthesizing or after the run finishes. source: anyOf: - type: string - type: 'null' title: Source panel: anyOf: - type: string - type: 'null' title: Panel description: >- Canonical panel id: wide, consensus, or seneca. A stored Quick Pulse run still reports quick_pulse. prompt: type: string title: Prompt credits_reserved: type: integer title: Credits Reserved description: >- Credits reserved when the run was accepted. This is the panel price and does not change if the run is later refunded. credits_charged: anyOf: - type: integer - type: 'null' title: Credits Charged description: >- Null while the run is in progress. Equal to credits_reserved when a customer report is delivered, including a partial run that has a report. 0 after a verified technical failure with no customer report; that refund restores the reserve. failure_code does not change between a charged failure and a refunded one. failure_code: anyOf: - type: string - type: 'null' title: Failure Code description: research_failed when status is failed. Null otherwise. created_at: type: string format: date-time title: Created At completed_at: anyOf: - type: string format: date-time - type: 'null' title: Completed At progress: $ref: '#/components/schemas/PublicProgressOut' additionalProperties: false type: object required: - id - status - prompt - credits_reserved - created_at - progress title: PublicRunOut PublicErrorResponse: properties: error: $ref: '#/components/schemas/PublicErrorObject' additionalProperties: false type: object required: - error title: PublicErrorResponse description: Error envelope returned by every /v1 failure. Documentation model only. PublicProgressOut: properties: total_models: type: integer title: Total Models completed_models: type: integer title: Completed Models successful_models: type: integer title: Successful Models failed_models: type: integer title: Failed Models additionalProperties: false type: object required: - total_models - completed_models - successful_models - failed_models title: PublicProgressOut PublicErrorObject: properties: code: type: string title: Code message: type: string title: Message details: additionalProperties: true type: object title: Details additionalProperties: false type: object required: - code - message title: PublicErrorObject securitySchemes: SenecaApiKey: type: http description: >- Workspace API key created while signed in on the Developers page. Send Authorization: Bearer seneca_sk_… A Clerk session token is not accepted. scheme: bearer ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.