⚡ Workflows
Scatter-Gather
Broadcast the same task to multiple independent solvers simultaneously, then score and select the best answer (or merge the best parts). Maximizes quality when single-agent output is unreliable or when diverse approaches improve coverage.
Bundle files
📄ORCHESTRATION.md— Orchestration
1# ORCHESTRATION.md
2
3## Diagram
4
5 ┌──────────────────────────────────────────┐
6 │ Dispatcher │
7 │ (broadcast same task to all solvers) │
8 └──────┬──────────┬──────────┬─────────────┘
9 │ │ │
10 ▼ ▼ ▼
11 ┌─────────┐ ┌─────────┐ ┌─────────┐
12 │Solver 1 │ │Solver 2 │ │Solver N │
13 │(attempt)│ │(attempt)│ │(attempt)│
14 └────┬────┘ └────┬────┘ └────┬────┘
15 │ │ │
16 └────────────┼────────────┘
17 ▼
18 ┌──────────────────────────────────────────┐
19 │ Aggregator │
20 │ (score + select best / merge results) │
21 └──────────────────────────────────────────┘
22
23
24## Workflow
25Scatter-Gather
26
27## Objective
28Broadcast the same task to multiple independent solvers simultaneously, then score and select the best answer (or merge the best parts). Maximizes quality when single-agent output is unreliable or when diverse approaches improve coverage.
29
30## Roles
31- **Dispatcher:** formulates the task and broadcasts it identically to all solvers.
32- **Solvers (N agents):** each independently attempts the full task using their own approach, persona, or model.
33- **Aggregator:** receives all responses, scores them against defined criteria, and returns the winner or a synthesized composite.
34
35## When to use
36- High-stakes outputs where quality matters more than speed (critical copy, architecture decisions, legal drafts).
37- Tasks where diversity of approach improves coverage (red-flag detection, research synthesis).
38- When a single agent is likely to miss edge cases or make systematic errors.
39- A/B testing different agent configurations on real work.
40
41## Protocol
421. Dispatcher defines the task with explicit output requirements and scoring criteria.
432. Task is broadcast to all N solvers simultaneously.
443. All solvers produce independent responses without seeing each other's work.
454. Aggregator receives all N responses.
465. Aggregator scores each response using the criteria defined in step 1.
476. Aggregator selects best single response OR synthesizes a composite using strengths from each.
487. Aggregator produces final output + scoring rationale + dissent summary.
49
50## Scoring criteria (configurable)
51- Accuracy / correctness
52- Completeness (coverage of requirements)
53- Conciseness (no padding or redundancy)
54- Risk flags identified
55- Adherence to format/constraints
56
57## Rules
58- Solvers must be isolated — no cross-contamination of outputs during generation.
59- Dispatcher task brief must be identical for all solvers.
60- Aggregator must document why the winner was selected (or how the composite was built).
61- Minimum N = 2; recommended N = 3 for most tasks.
62- If all solvers produce low-confidence outputs, escalate rather than picking the least-bad one.
63
64## Deliverables
65- Final selected or composite answer.
66- Scoring matrix for all N responses.
67- Dissent log: meaningful differences between solvers worth noting.
68- Recommended N for similar future tasks.
69
Lines: 69 | Words: 403
Install
Copy bundle to your OpenClaw workspace.
curl -fsSL https://raw.githubusercontent.com/cerealskill/openclaw-agents/main/install.sh | bash -s workflow scatter-gather ENRate this agent
Loading...
Sign in to rate this agent
Includes
- ✓ ORCHESTRATION.md
Info
- Slug
- scatter-gather
- Lines
- 69
- Words
- 403