Generating a Task List from Requirements

Workflow

Step 1: Project Detection

Explicitly detect and list the project's conventions before generating tasks:

1. Test Command Detection — Use the lookup table below to resolve the concrete command.
2. Directory Detection — Check for common source dirs (src/, lib/, app/) and test dirs (tests/, test/, spec/, __tests__/).
3. Documentation Tool Detection — Identify the documentation tool by checking package.json, pyproject.toml, requirements.txt, or a Doxyfile for references to tools such as typedoc, jsdoc, sphinx, mkdocs, or doxygen.

Test Command Lookup Table

Step 2: Validation Checkpoint

Verify the detected test command works before proceeding.

Run the detected command. If it fails, ask the user to confirm: "Detected test command: [command]. Is this correct?" Only proceed if tests execute successfully.

Step 3: Requirements Analysis

Break down the feature/PRD into implementation tasks:

1. Identify User-Visible Behaviors

2. Apply First-Slice Heuristics — Choose the smallest slice that proves behavior:

   Project Type	First Slice
   Web App / API Service	Request/integration test
   Service / Domain logic	Service or unit test
   Background Job	Job/worker test
   External Integration	Client/fetcher layer test
   Library / SDK	Public API test

3. Create Parent Task Groups — Group related behaviors with TDD quadruplets.

Step 4: Generate Task List

Create /tasks/tasks-[feature-name].md using the structure below.

Task File Template

Each parent group follows the TDD quadruplet: write failing test → confirm fail → implement → confirm pass.

# Tasks: [Feature Name]

## Relevant Files
- `src/[module]/[file].[ext]` — [brief description]
- `tests/[module]/[file].test.[ext]` — [brief description]

## Tasks

- [ ] 0.0 Create feature branch `feat/[feature-name]`

- [ ] 1.0 [Parent behavior group]
  - [ ] 1.1 Write failing test for [smallest behavior slice]
  - [ ] 1.2 Run `[test command]` — confirm test 1.1 fails
  - [ ] 1.3 Implement [feature code] to make test 1.1 pass
  - [ ] 1.4 Run `[test command]` — confirm all tests pass

- [ ] 2.0 [Next parent behavior group]
  - [ ] 2.1 Write failing test for ...
  - [ ] 2.2 Run — confirm fail
  - [ ] 2.3 Implement ...
  - [ ] 2.4 Run — confirm pass

Step 5: Final Validation

1. Review Generated Tasks

   • Are all user behaviors covered?
   • Is the TDD quadruplet pattern consistent?
   • Are file paths correct for detected directories?

2. Verify Test Integration

   • Test command references are correct.
   • File paths match project structure.