Split this across agents

What it asks you for

The jobrequired

The whole piece of work.

Workers

How many to run at once.

What it tells the AI to do

The full instructions, exactly as they run.

Split this across {{ workers | default(3) }} sub-agents:

{{ job }}

Only split work that is genuinely INDEPENDENT. Two agents editing the same file,
or one waiting on another's conclusion, is slower than doing it yourself plus a
merge conflict.

1. Decompose into pieces with no shared state, and say what each returns.
2. Give each sub-agent the full context it needs. A worker that has to guess the
   surrounding decisions produces work you then have to redo.
3. Run them, then MERGE deliberately: resolve contradictions rather than
   concatenating, and name the ones you resolved.

Report what each worker found, then the merged result. If two workers disagreed,
that disagreement is a finding, not a formatting problem.

• agent
• delegate
• parallel
• subagent

More agents skills

• Create an agentDesign a reusable agent with a real job and real limits.
• Fix an agent's behaviourDiagnose why an agent misbehaves and correct its instructions.