xingyue
522afdd4bd
- Install @changesets/cli with fixed mode (all @uncaged/* packages sync version) - Fix package exports: add bun condition, point import to dist/ - Bump all packages to 0.4.0 via changeset version - Auto-generated CHANGELOG.md for each package - Ignore workflow-dashboard (private) - Add npm scripts: changeset, version, release - publish.sh: support workspace:^ prefix matching Closes #241, Closes #242
@uncaged/workflow-agent-llm
AgentFn adapter that calls an OpenAI-compatible POST /chat/completions endpoint using LlmProvider from @uncaged/workflow-runtime.
Single-turn: system text is the current role’s systemPrompt, user text is the thread’s initial prompt (ctx.start.content). Errors from HTTP, JSON, or empty choices are thrown as Error with a JSON payload string.
Install
bun add @uncaged/workflow-agent-llm @uncaged/workflow-runtime zod
In this monorepo: "@uncaged/workflow-agent-llm": "workspace:*", "@uncaged/workflow-runtime": "workspace:*" (and satisfy zod ^4 as required by @uncaged/workflow-runtime).
Usage
import { createLlmAdapter } from "@uncaged/workflow-agent-llm";
const agent = createLlmAdapter({
baseUrl: "https://api.openai.com/v1",
apiKey: process.env.OPENAI_API_KEY!,
model: "gpt-4.1-mini",
});
API overview
| Export | Description |
|---|---|
createLlmAdapter(provider) |
LlmProvider → AgentFn |
chatCompletionText({ provider, messages }) |
Low-level Result<string, LlmChatError> helper |
LlmMessage |
{ role: "system" | "user" | "assistant"; content: string } |
LlmChatError |
Discriminated error kinds for failed completions |