Skip to main content

Documentation Index

Fetch the complete documentation index at: https://gump.build/docs/llms.txt

Use this file to discover all available pages before exploring further.

​
Choosing the Right Agent

Different agents have different strengths. Picking the right one for each step is the main lever for optimizing cost and quality.

​
The trade-off

More capable agents cost more and are slower, but succeed more often on the first try. Cheaper agents are fast and affordable, but may need retries or escalation. Gump’s retry system lets you start cheap and pay more only when needed.

​
Rules of thumb

​
Planning (type: split)

Use a strong agent. The plan determines everything — a bad decomposition wastes all downstream work. Claude Opus or Claude Sonnet are good choices.

​
Implementation (type: code)

Start cheap. Qwen, Claude Haiku, or OpenCode handle most implementations if the plan is good. Use escalation in retry for harder cases. 
retry:
  - attempt: 3
    agent: claude-haiku
  - attempt: 5
    agent: claude-sonnet
  - attempt: 7
    agent: claude-opus
  - exit: 8

​
Review (type: validate, as gate)

Use a strong agent with session: new. The reviewer should be at least as capable as the implementer. Using a different provider (e.g., Gemini for review when Claude implemented) reduces shared blind spots.

​
Mixing agents

Gump is agent-agnostic. Mix agents freely: 
steps:
  - name: decompose
    type: split
    run:
      agent: claude-opus
    ...
    each:
      - name: impl
        type: code
        run:
          agent: qwen
        retry:
          - attempt: 3
            agent: claude-sonnet
          - exit: 5
The only constraint: session: from: <step> requires both steps to use the same agent.

​
Let the data decide

After a few runs, use gump report to see which agents succeed on which types of tasks. This is the feedback loop: run → measure → adjust → run.