// Use when executing a single bead/task with test-driven development approach
Use when starting new feature work to turn ideas into fully formed designs through collaborative dialogue
Use after implementation to review code against spec and quality standards
Use when an implementation plan exists and needs to be broken into parallelizable work units with dependencies
Use when bead implementation is complete and ready for integration
Use when a design document exists and needs to be broken into detailed implementation tasks
| name | subagent-dev |
| description | Use when executing a single bead/task with test-driven development approach |
Execute a single bead using test-driven development. Fresh context, focused work, clean commits.
Understand what needs to be built:
Before writing any code, clarify ambiguities:
Don't proceed with assumptions - get clarity upfront.
# Create test file if needed
# Write test that captures the requirement
# Run to verify it fails
Test must fail for the right reason - "function not defined" or "assertion failed", not syntax error.
Write the simplest code that makes the test pass:
# Run the specific test
# Verify it passes
# Run related tests to check for regressions
Before committing, check:
git add -A
git commit -m "feat: <what was implemented>
- <key detail 1>
- <key detail 2>"
Never:
If stuck: