› cavi-ai/
GitHub ↗

Codex (OpenAI) operations

This client mirrors and verifies upstream-compatible behavior. Upstream runtimes remain the canonical protocol owners.

Runtime-only provider over the OpenAI Responses API. Auth: Authorization: Bearer <apiKey>. providerKind codex-responses, protocolVersion responses-v1. Default model gpt-5-codex (CODEX_DEFAULT_MODEL) when a run omits one. Background responses make it stateful — getRun/cancelRun are implemented. Supports the batch surface over the OpenAI Batch API.

Capability (CODEX_RUNTIME_SUPPORT): runs ✅ · getRun ✅ · cancelRun ✅ · streamRun ✅ · batch ✅.

startRun

Signature
client.startRun(body: RuntimeRunStartBody): Promise<RuntimeRunStatus>
HTTP
POST /v1/responses
Capability
supports.runs

Maps the universal body to an OpenAI Responses request; output/tokens are normalized from the response. See runtime · startRun for the full field tables.

Example

ts
import { createCodexProviderModule } from "@cavi-ai/api-client/providers/codex";
// … construct the runtime client, then:
const run = await client.startRun({ input: "Hi", model: "gpt-5-codex" });

getRun

Signature
client.getRun(runId: string): Promise<RuntimeRunStatus>
HTTP
GET /v1/responses/:id
Capability
supports.runs

Reads a background response by id. Field tables per runtime · getRun.

cancelRun

Signature
client.cancelRun(runId: string): Promise<{ status: string }>
HTTP
POST /v1/responses/:id/cancel
Capability
supports.runs

Cancels an in-flight background response. See runtime · cancelRun.

streamRun

Signature
client.streamRun(body: RuntimeRunStartBody, handlers: RunEventStreamHandlers): Promise<void>
HTTP
POST /v1/responses (SSE)
Capability
supports.streaming

Streams the Responses SSE event feed, normalized to canonical run-stream events. See runtime · streamRun.

submitBatch / getBatch / cancelBatch / getBatchResults

HTTP
POST /v1/batches · GET /v1/batches/:id · POST /v1/batches/:id/cancel · GET /v1/files/:id/content
Capability
supports.batch

submitBatch uploads a JSONL input file (POST /v1/files) and creates a batch against the /v1/responses endpoint; getBatchResults reads the batch (GET /v1/batches/:id) then downloads its output/error file (GET /v1/files/:id/content). Field tables per runtime · batch operations.