openserv-agent-sdk
oleh issa-me-sush
Coding Agents & IDEsRingkasan
Build and deploy autonomous AI agents using the OpenServ SDK (@openserv-labs/sdk)
Deskripsi
Apa yang Dilakukan Skill Ini
A TypeScript SDK for building and deploying AI agents on the OpenServ platform. Agents expose capabilities that the platform can invoke via webhooks, cron schedules, or x402 micropayments. LLM calls can be delegated to the platform (no API key needed) or handled by any provider you supply.
Agents can run LLM calls through the platform without managing API keys or billing, lowering the setup cost compared to self-hosted LLM pipelines.
Kapan Menggunakannya
- Building a paid API endpoint that wraps a third-party data source
- Scheduling a daily content generation job triggered by cron
- Deploying a research agent invoked by other agents in a multi-agent workflow
- Exposing a data transformation tool callable via x402 micropayment
- Registering an on-chain discoverable agent identity on Base mainnet
Contoh Alur Kerja
User asks: Deploy an agent that generates haiku poems for paying callers
- 1
Install @openserv-labs/sdk, @openserv-labs/client, and zod
- 2
Create an Agent instance with a system prompt and add a runless capability named generate_haiku with a description
- 3
Call provision() with a workflow name, descriptive goal, and an x402 trigger set to the desired price and a 600-second timeout
- 4
Pass agent.instance to provision() so credentials are written to .env automatically
- 5
Call run(agent) to start the HTTP server and connect to the platform proxy
A live haiku generator agent registered on OpenServ, accepting x402 micropayment requests and returning poems without requiring any LLM API key
Persyaratan
- OpenServ platform account (auto-created by provision() on first run, but a wallet private key is stored locally)
- ETH on Base mainnet to fund on-chain ERC-8004 agent identity registration
- OPENAI_API_KEY if using the process() method for direct OpenAI calls
Instalasi
clawhub install issa-me-sush/openserv-agent-sdkJalankan perintah CLI-nya, atau tempel prompt ke Claude Code, Cursor, Codex, atau coding agent apa pun yang Anda pakai.