Back to catalog / Agent workflows
Break an open-ended writing task into parallel drafts
Let an orchestrator choose task-specific angles and specialized workers produce comparable drafts.
Goal
Generate distinct candidate drafts for one brief for an editor to compare.
Steps
- Define an orchestrator XML task list using
<task><type>...and cap it at 2–3 distinct angles. [orchestrate] - Parse and validate each worker’s type/description, then call
llm_call(prompt, system_prompt, model)for separate drafts. [orchestrate] - Present results side by side with the original brief and angle labels; check for duplication and factual conflicts before editorial selection. [orchestrate]
How to verify
Workers produce meaningfully different angles; malformed task plans stop visibly instead of dropping drafts silently.
Troubleshooting
If XML parsing fails, validate tags and request one bounded retry; route repeated failure to an editor.
Permissions
Prerequisites
- Python 3.11+, anthropic SDK, an API key, and an approved product brief.
Permissions
- Provide public or approved product facts only; workers draft copy without publishing it.
Limitations
The notebook uses a simple XML/regex example, not a general secure parser; do not publish generated copy without editorial review.
Pinned upstream source
- Repository
- anthropics/anthropic-cookbook
- File path
- patterns/agents/orchestrator_workers.ipynb
- Pinned commit
9c1e493ddd2b82df399d2a61b9c806c539e18be6- Source checked
- 2026-10-14 · Reviewed against the pinned upstream source
Repository metrics
These values describe the upstream repository, not popularity or likes for an individual skill or playbook.
- Stars
- Not collected
- Forks
- Not collected
- Observed at
- Not collected
No metrics snapshot is available yet.
License: MIT
Main topic
Turn ecommerce campaign materials into a reviewable weekly brief