Writing Plans

Overview

Write comprehensive implementation plans assuming the engineer has zero context for our codebase and questionable taste. Document everything they need to know: which files to touch for each task, code, testing, docs they might need to check, how to test it. Give them the whole plan as bite-sized tasks. DRY. YAGNI. TDD. Frequent commits.

Assume they are a skilled developer, but know almost nothing about our toolset or problem domain. Assume they don't know good test design very well.

Announce at start: "I'm using the writing-plans skill to create the implementation plan."

Context: This should be run in a dedicated worktree (created by brainstorming skill).

Save plans to: docs/plans/YYYY-MM-DD-<feature-name>.md

Bite-Sized Task Granularity

Each step is one action (2-5 minutes):

• "Write the failing test" - step
• "Run it to make sure it fails" - step
• "Implement the minimal code to make the test pass" - step
• "Run the tests and make sure they pass" - step
• "Commit" - step

Plan Document Header

Every plan MUST start with this header:

# [Feature Name] Implementation Plan

> **For Claude:** REQUIRED SUB-SKILL: Use superpowers:executing-plans to implement this plan task-by-task.

**Goal:** [One sentence describing what this builds]

**Architecture:** [2-3 sentences about approach]

**Tech Stack:** [Key technologies/libraries]

## Alternatives

### [Decision: e.g. State Management Approach]

| Criteria | Approach A: [Name] | Approach B: [Name] |
|---|---|---|
| Performance | ... | ... |
| Complexity | ... | ... |
| Maintainability | ... | ... |

**Decision:** [Chosen approach] — [One-line rationale]

---

Alternatives Exploration

Every non-trivial decision in the plan MUST include at least two approaches with trade-off analysis.

What counts as non-trivial:

• Architecture choices (e.g. where to put new code)
• Data flow design (e.g. state management approach)
• API design (e.g. endpoint structure, payload shape)
• Testing strategy (e.g. unit vs integration)

What does NOT need alternatives:

• File naming that follows existing conventions
• Import statements
• Trivial implementation details

Format for each decision:

### [Decision: Short Description]

| Criteria | Approach A: [Name] | Approach B: [Name] |
|---|---|---|
| Performance | ... | ... |
| Complexity | ... | ... |
| Maintainability | ... | ... |

**Decision:** [Chosen approach] — [One-line rationale]

Rules:

• Minimum 2 approaches per non-trivial decision
• Always include performance, complexity, and maintainability rows
• Add extra criteria rows when relevant (e.g. security, testability, migration cost)
• State the chosen approach with a clear rationale
• If only one viable approach exists, explain why alternatives were ruled out

API Verification

When the plan references external APIs, SDKs, or protocols:

1. Document every API assumption with its source URL
2. Create an "API Assumptions" table in the plan:

1. All assumptions must be verified before ACT phase
2. WebFetch/WebSearch the official docs — never rely on memory or guesses

Why This Matters

Unverified API assumptions propagate into implementation, causing features to be unimplementable as designed. Verification at PLAN time costs minutes; rework at ACT time costs hours.

Task Structure

### Task N: [Component Name]

**Files:**
- Create: `exact/path/to/file.py`
- Modify: `exact/path/to/existing.py:123-145`
- Test: `tests/exact/path/to/test.py`

**Step 1: Write the failing test**

def test_specific_behavior(): result = function(input) assert result == expected

**Step 2: Run test to verify it fails**

Run: `pytest tests/path/test.py::test_name -v`
Expected: FAIL with "function not defined"

**Step 3: Write minimal implementation**

def function(input): return expected

**Step 4: Run test to verify it passes**

Run: `pytest tests/path/test.py::test_name -v`
Expected: PASS

**Step 5: Commit**

git add tests/path/test.py src/path/file.py git commit -m "feat: add specific feature"

Remember

• Exact file paths always
• Complete code in plan (not "add validation")
• Exact commands with expected output
• Reference relevant skills with @ syntax
• DRY, YAGNI, TDD, frequent commits

Self-Review Gate

Before submitting the plan, verify ALL items pass:

• [ ] All file paths verified (existing files confirmed, new files marked as Create:)
• [ ] Steps are bite-sized (2-5 minutes each, one action per step)
• [ ] Risks identified with mitigation (what could go wrong and how to handle it)
• [ ] TDD applied where appropriate (test-first for core logic, test-after for UI)
• [ ] No over-engineering (YAGNI check — remove anything not directly needed)

If any item fails: Fix the plan before proceeding. Do not hand off an incomplete plan.

Execution Handoff

After saving the plan, offer execution choice:

"Plan complete and saved to `docs/plans/<filename>.md`. Two execution options:

1. Subagent-Driven (this session) - I dispatch fresh subagent per task, review between tasks, fast iteration

2. Parallel Session (separate) - Open new session with executing-plans, batch execution with checkpoints

Which approach?"

If Subagent-Driven chosen:

• REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development
• Stay in this session
• Fresh subagent per task + code review

If Parallel Session chosen:

• Guide them to open new session in worktree
• REQUIRED SUB-SKILL: New session uses superpowers:executing-plans