@radzor/llm-completion

Universal LLM completion client supporting OpenAI, Anthropic, and Ollama. Provides chat completions, streaming, conversation history, and automatic retries with exponential backoff.

AI & MLv0.1.0typescriptpythonServerllmopenaianthropicollamachatstreamingaicompletionby Radzor

Install

View source on GitHub →

$npx radzor@latest add llm-completion

⚠ Constraints: Requires an API key for OpenAI and Anthropic. Ollama requires a local instance running. Uses the native fetch API (Node.js 18+ or browser).

Inputs

Name	Type	Default	Description
provider*	'openai' \| 'anthropic' \| 'ollama'	—	LLM provider to use.
apiKey	string	—	API key for the provider. Required for openai and anthropic, optional for ollama.OPENAI_API_KEY
model*	string	—	Model identifier (e.g. gpt-4o, claude-sonnet-4-20250514, llama3).
baseUrl	string	—	Override the API base URL. Useful for proxies or self-hosted instances.
maxTokens	number	4096	Maximum tokens to generate in the response.
temperature	number	0.7

{ "llm": { "constraints": "Requires an API key for OpenAI and Anthropic. Ollama requires a local instance running. Uses the native fetch API (Node.js 18+ or browser).", "usageExamples": "llm/examples.md", "integrationPrompt": "llm/integration.md" }, "name": "@radzor/llm-completion", "tags": [ "llm", "openai", "anthropic", "ollama", "chat", "streaming", "ai", "completion" ], "events": [ { "name": "onChunk", "payload": { "done": "boolean", "content": "string" }, "description": "Fired for each streamed token. done=true signals the stream is complete." }, { "name": "onComplete", "payload": { "model": "string", "content": "string", "finishReason": "string" }, "description": "Fired when a completion (streamed or non-streamed) finishes." }, { "name": "onError", "payload": { "code": "string", "message": "string", "provider": "string" }, "description": "Fired when an API call fails after all retries." } ], "inputs": [ { "name": "provider", "type": "'openai' | 'anthropic' | 'ollama'", "required": true, "description": "LLM provider to use." }, { "name": "apiKey", "type": "string", "envVar": "OPENAI_API_KEY", "required": false, "description": "API key for the provider. Required for openai and anthropic, optional for ollama." }, { "name": "model", "type": "string", "required": true, "description": "Model identifier (e.g. gpt-4o, claude-sonnet-4-20250514, llama3)." }, { "name": "baseUrl", "type": "string", "required": false, "description": "Override the API base URL. Useful for proxies or self-hosted instances." }, { "name": "maxTokens", "type": "number", "default": 4096, "required": false, "description": "Maximum tokens to generate in the response." }, { "name": "temperature", "type": "number", "default": 0.7, "required": false, "description": "Sampling temperature (0.0 = deterministic, 1.0 = creative)." }, { "name": "systemPrompt", "type": "string", "required": false, "description": "System prompt prepended to all conversations." } ], "radzor": "1.0.0", "actions": [ { "name": "complete", "params": [ { "name": "prompt", "type": "string", "description": "The user message to send." } ], "returns": "Promise<CompletionResult>", "description": "Send a prompt and get a complete response. Adds to conversation history." }, { "name": "stream", "params": [ { "name": "prompt", "type": "string", "description": "The user message to send." } ], "returns": "Promise<string>", "description": "Send a prompt and stream the response token by token via onChunk events." }, { "name": "getHistory", "params": [], "returns": "ChatMessage[]", "description": "Get the full conversation history." }, { "name": "clearHistory", "params": [], "returns": "void", "description": "Clear conversation history (preserves system prompt)." }, { "name": "setSystemPrompt", "params": [ { "name": "prompt", "type": "string", "description": "The new system prompt." } ], "returns": "void", "description": "Set or update the system prompt for future completions." } ], "outputs": [ { "name": "completionResult", "type": "CompletionResult", "fields": [ { "name": "content", "type": "string" }, { "name": "model", "type": "string" }, { "name": "usage", "type": "{ promptTokens: number, completionTokens: number, totalTokens: number }", "description": "Token usage statistics for the completion." }, { "name": "finishReason", "type": "string" } ], "description": "The full completion response including content, usage stats, and finish reason." } ], "runtime": "server", "version": "0.1.0", "category": "ai", "languages": [ "typescript", "python" ], "description": "Universal LLM completion client supporting OpenAI, Anthropic, and Ollama. Provides chat completions, streaming, conversation history, and automatic retries with exponential backoff.", "dependencies": { "packages": {} }, "composability": { "connectsTo": [ { "output": "completionResult", "mapField": "content", "compatibleWith": [ "@radzor/text-to-speech.action.synthesize.text", "@radzor/structured-output.action.extract.text", "@radzor/markdown-render.action.render.markdown" ] }, { "output": "completionResult", "mapField": "content", "description": "Send LLM-generated text to any notification channel.", "compatibleWith": [ "@radzor/email-send.action.send.message", "@radzor/slack-bot.action.sendMessage.text", "@radzor/discord-bot.action.sendMessage.content", "@radzor/telegram-bot.action.sendMessage.text", "@radzor/sms-send.action.send.body", "@radzor/push-notification.action.sendToDevice.notification" ] }, { "output": "completionResult", "mapField": "content", "description": "Use LLM output as image generation prompt.", "compatibleWith": [ "@radzor/image-generation.action.generate.prompt" ] }, { "output": "completionResult", "mapField": "content", "description": "Persist or render LLM output.", "compatibleWith": [ "@radzor/pdf-generate.action.fromHtml.html", "@radzor/search-index.action.index.documents", "@radzor/embeddings-store.action.add.text" ] } ] } }

@radzor/llm-completion

Install

Inputs

Outputs

Actions

Events

Composability

radzor.manifest.json

Version History