Write custom CoordinatorStrategy for code review routing

[sunilp]

What

Create a custom CoordinatorStrategy that routes code review tasks to the most appropriate reviewer agent based on programming language, diff complexity, and reviewer expertise.

Why

The examples/custom-coordinator-strategy/ currently has a healthcare example. A code review routing strategy would be relatable to most developers and demonstrate the Coordinator pattern in a familiar domain.

How

1. Subclass DefaultCoordinatorStrategy (see sdk/python/jamjet/coordinator/default_strategy.py)
2. Override score() to weight language expertise and diff complexity
3. Create mock agents (e.g., python-reviewer, rust-reviewer, frontend-reviewer)
4. Add as examples/code-review-routing/main.py

Key Concepts

from jamjet.coordinator import DefaultCoordinatorStrategy, DimensionScores

class CodeReviewStrategy(DefaultCoordinatorStrategy):
    async def score(self, task, candidates, weights, context):
        # Weight language match heavily
        # Consider diff size for latency estimation
        ...

Resources

• examples/custom-coordinator-strategy/main.py — healthcare strategy example
• sdk/python/jamjet/coordinator/strategy.py — base class and types