Autonomy direction

One agent or many?

A multi-agent environment is not automatically smarter. It is a coordination design: who plans, who specializes, what information crosses boundaries, and which actions require approval.

Single agent

One context, one owner

Good when the task is coherent and the agent needs to see the whole problem. Fewer handoff failures and simpler debugging.

Manager + specialists

Central coordination

A manager keeps the conversation and asks focused specialists for research, coding, testing, or review. The manager assembles the result.

Handoffs

Explicit ownership transfer

A specialist takes over a phase or conversation. Useful when a new context is cleaner, but boundaries and state must be designed.

A safe autonomy ladder

1

Observe

Read files, inspect state, and produce a plan. No external side effects.

2

Suggest

Draft commands, patches, messages, or decisions for a human to review.

3

Act reversibly

Make bounded edits, run tests, or create artifacts with a clear diff and undo path.

4

Coordinate

Delegate narrow tasks in parallel, then require a synthesis and evidence check.

5

Operate

Allow recurring or external actions only with scope, budgets, logs, approval gates, and a stop mechanism.

Design the interfaces between agents

InterfaceMake explicitGood handoff result
TaskObjective, scope, exclusions, success criteriaA specialist knows what done means
ContextRelevant files, facts, assumptions, unknownsNo need to rediscover the entire project
OutputFormat, evidence, changed paths, open questionsThe next agent can verify and continue
AuthorityRead, write, network, spending, or messaging permissionsSide effects happen only inside the intended boundary
FailureTimeout, retry, escalation, and human stop behaviorPartial work is visible instead of silently lost

Good first experiments

Parallel review

Ask separate reviewers to inspect accessibility, security, and test coverage. Give each the same commit or snapshot, then compare findings and resolve conflicts centrally.

Research then build

Let one agent collect current primary sources and another turn only those sources into an implementation outline. Keep the source packet visible.

Generator then critic

Have one agent draft an artifact and a second check it against a fixed rubric. The critic should report evidence and failures, not merely a confidence score.

Watch with a stop button

For recurring work, begin with read-only monitoring and meaningful-change alerts. Add mutation only after false positives, retries, and costs are understood.

When not to add another agent

If the task is unclear, the repository is unmapped, the tools are unsafe, or the output cannot be evaluated, another agent adds more context and more failure surfaces. Improve the brief, interfaces, and proof first.

Review agent-orchestration sources →