#The docs are unclear about how to use lobster with llm-task

1 messages · Page 1 of 1 (latest)

sterile dagger
#

The OpenClaw docs and the Lobster Github docs don't seem to match up.

  1. When I need an LLM in a Lobster pipeline, should I be using llm.invoke or llm-task.invoke?

  2. Do I need to pass the OpenClaw gateway URL to Lobster if I'm running it from an OpenClaw agent? Why?

  3. Why is it described as a "Native" pipeline step?

  4. Why is there also the option to call openclaw.invoke --tool llm-task; is there a meaningful difference? Is it redundant?

  5. In the OpenClaw does, it says you should In OpenClaw tool calls, set pipeline to the file path. What does this mean?

My agent is having problems getting calls from Lobster to work, and I'm unable to help them.

frozen hawkBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

sterile dagger
#

So that is upsetting.