Coordination Patterns#
Source: examples/patterns/coordination_patterns.py
Introduction#
Blackboard-system architecture motivates shared-state collaboration among specialized problem solvers, AutoGen informs practical multi-agent implementation choices, and Human-AI collaboration by design clarifies governance value in shared workspace reasoning. This example compares round-based coordination and blackboard-specialized runs with explicit execution records.
Technical Implementation#
Configure
Tracerwith JSONL + console output so each run emits machine-readable traces and lifecycle logs.Build the runtime surface (public APIs only) and execute
RoundBasedCoordinationPattern.run(...)with a fixedrequest_id.Capture structured outputs from runtime execution and preserve termination metadata for analysis.
Print a compact JSON payload including
trace_infofor deterministic tests and docs examples.
flowchart LR
A["Input prompt or scenario"] --> B["main(): runtime wiring"]
B --> C["RoundBasedCoordinationPattern.run(...)"]
C --> D["blackboard workers contribute and aggregate shared state"]
C --> E["Tracer JSONL + console events"]
D --> F["ExecutionResult/payload"]
E --> F
F --> G["Printed JSON output"]
1from __future__ import annotations
2
3import json
4from pathlib import Path
5
6import design_research_agents as drag
7
8
9def _summarize(result: drag.ExecutionResult) -> dict[str, object]:
10 return result.summary()
11
12
13def main() -> None:
14 """Run one round-based coordination and one blackboard pass."""
15 tracer = drag.Tracer(
16 enabled=True,
17 trace_dir=Path("artifacts/examples/traces"),
18 enable_jsonl=True,
19 enable_console=True,
20 )
21
22 with drag.LlamaCppServerLLMClient(context_window=16384) as llm_client:
23 peer_a = drag.DirectLLMCall(llm_client=llm_client, tracer=tracer)
24 peer_b = drag.DirectLLMCall(llm_client=llm_client, tracer=tracer)
25
26 # Split ids by pattern variant to keep networked and blackboard traces distinct.
27
28 coordination_request_id = "example-workflow-round-based-coordination-design-001"
29 coordination = drag.RoundBasedCoordinationPattern(
30 peers={
31 "peer_b": peer_b,
32 "peer_a": peer_a,
33 },
34 max_rounds=1,
35 tracer=tracer,
36 )
37 coordination_result = coordination.run(
38 "Exchange one concise proposal for a field-serviceable sensor enclosure.",
39 request_id=coordination_request_id,
40 )
41
42 # Split ids by pattern variant to keep networked and blackboard traces distinct.
43
44 blackboard_request_id = "example-workflow-blackboard-design-001"
45 blackboard = drag.BlackboardPattern(
46 peers={
47 "peer_b": peer_b,
48 "peer_a": peer_a,
49 },
50 max_rounds=1,
51 stability_rounds=1,
52 tracer=tracer,
53 )
54 blackboard_result = blackboard.run(
55 "Compare two concept options and make one concise serviceability recommendation.",
56 request_id=blackboard_request_id,
57 )
58
59 print(
60 json.dumps(
61 {
62 "blackboard": _summarize(blackboard_result),
63 "round_based_coordination": _summarize(coordination_result),
64 },
65 ensure_ascii=True,
66 indent=2,
67 sort_keys=True,
68 )
69 )
70
71
72if __name__ == "__main__":
73 main()
Expected Results#
Run Command
PYTHONPATH=src python3 examples/patterns/coordination_patterns.py
Example output shape (values vary by run):
{
"round_based_coordination": {
"success": true,
"final_output": "<example-specific payload>",
"terminated_reason": "<string-or-null>",
"error": null,
"trace": {
"request_id": "<request-id>",
"trace_dir": "artifacts/examples/traces",
"trace_path": "artifacts/examples/traces/run_<timestamp>_<request_id>.jsonl"
}
},
"blackboard": {
"success": true,
"final_output": "<example-specific payload>",
"terminated_reason": "<string-or-null>",
"error": null,
"trace": {
"request_id": "<request-id>",
"trace_dir": "artifacts/examples/traces",
"trace_path": "artifacts/examples/traces/run_<timestamp>_<request_id>.jsonl"
}
}
}