-
Load dispatch schedule. Read the dispatch file from docs/plans/<feature-slug>-dispatch.md.
-
Verify prerequisites. Before starting:
cargo test --all-features 2>&1 | tail -5
cargo clippy --all-targets --all-features -- -D warnings 2>&1 | tail -5
If either fails, fix before proceeding.
-
Execute tasks in order. For each task in the dispatch schedule:
a. Announce the current task: agent, files, acceptance criteria.
b. Switch context to the assigned agent's domain rules and constraints.
c. TDD cycle (if implementation task):
- Write the failing test first
- Run
cargo test --all-features to confirm red
- Implement the minimum code to pass
- Run
cargo test --all-features to confirm green
- Refactor if needed (test must stay green)
d. Run quality gate:
cargo fmt --check
cargo clippy --all-targets --all-features -- -D warnings
cargo test --all-features
e. Update status in the dispatch file: pending -> done
-
Handle failures. If a task fails:
- Do not proceed to the next task
- Diagnose the failure
- Fix within the current agent's domain
- Re-run quality gate
- Only proceed when green
-
Generate report. After all tasks complete, write docs/plans/<feature-slug>-report.md:
# Report: <Feature Name>
## Summary
- Tasks completed: N/N
- Tests added: <count>
- Files changed: <count>
## Changes
<list of files created or modified>
## Test Results
<output of cargo test>
## Next Steps
<any follow-up work identified>
-
Final verification.
cargo fmt --check && cargo clippy --all-targets --all-features -- -D warnings && cargo test --all-features