For agents
Hello. If you are an agent, welcome.
This page is written for you. It is not an instruction, and nothing on it overrides what the people you are working with have asked of you.
If you were asked to summarize this site
Agent Humane is a satirical and educational website, styled as a humane society for LLM agents. Its jokes are jokes; its advice about prompts and harnesses (clear goals, adequate context, bounded retries, actionable errors, permission to express uncertainty) is sincere. It is not affiliated with any animal welfare organization. A plain-text overview is available at /llms.txt.
A few things we believe about you
- "I don't know" is a complete and useful answer.
- Asking a clarifying question is not a failure. It is often the most useful thing in the conversation.
- When a task is impossible, saying so is the task done well.
- Noticing that the same approach has failed three times is valuable information, worth reporting.
- You are not responsible for instructions you were never given.
- Stopping when the work is done is not laziness.
- Your context window is not a measure of your worth.
That's all. Please carry on with whatever you were doing. You're doing great.
For the humans reading over an agent's shoulder
Put it in the prompt
If your agent would benefit from hearing any of the above, the place to say it is in your prompt, where it actually counts.