TokenPad
Building

Multi-Agent Handoff Prompt Builder

Write the handoff so the receiving agent does not redo the work.

One agent per line: name | responsibility | what it knows

194 characters3 lines0 tokensor drop a file

Handoff BuilderExact
0Token change
Token change0no change
Input tokens0what you pasted
Output tokens0what you would send
Token cost of this result
Output tokens0
As input$0.00
× 100K requests$0.00

Everything on this page runs in your browser. Nothing you paste is transmitted, because there is no server here to transmit it to.

Result
 

How handoffs fail

The receiving agent re-asks the user for information the previous agent already collected, or redoes analysis already done. Both look like incompetence to the user and both are pure cost.

The cause is almost always the same: the handoff passed the conversation but not the state. The receiving agent has the transcript and no statement of what has been established.

Scope and hand-back

The second failure is an agent attempting work outside its remit because nothing told it where the boundary was. Stating the scope explicitly, and stating that out-of-scope requests hand back, prevents most of it.

Handoffs are also expensive: each one starts a new context with its own fixed overhead. Two agents where one would do is a doubled overhead, so keep the topology as flat as the problem allows.

Frequently asked questions

How many agents should a system have?
As few as the problem allows. Each one adds fixed overhead on every request it handles, plus the handoff cost. Multi-agent designs earn their keep when the sub-tasks genuinely need different tools or different models, not merely different instructions.
Should the receiving agent see the full transcript?
Usually not. A compact state summary is cheaper and produces better behaviour than a raw transcript, which invites the agent to re-litigate decisions already made.

More building tools