Three solver agents tackle the same problem in parallel: a creative approach ("AI-Powered Adaptive System"), an analytical one ("Data-Driven Optimization Framework"), and a practical one ("Incremental Process Improvement"). A judge evaluates all three on cost, feasibility, and impact, then the winner is selected.
problem, criteria
|
v
+--- FORK_JOIN ---------+
| solver_1 | solver_2 | solver_3 |
+-----------+---------+
v
comp_judge_agent
v
comp_select_winner
Solver1Worker (comp_solver_1) -- approach: "creative", title: "AI-Powered Adaptive System".
Solver2Worker (comp_solver_2) -- approach: "analytical", title: "Data-Driven Optimization Framework".
Solver3Worker (comp_solver_3) -- approach: "practical", title: "Incremental Process Improvement".
JudgeAgentWorker (comp_judge_agent) -- Calls scoreSolution() for each solver on cost, feasibility, impact.
SelectWinnerWorker (comp_select_winner) -- Picks the highest-scoring solution from the judgment.
37 tests cover all three solvers, judging logic, and winner selection.
- RUNNING.md -- how to build and run this example