Writing Plans - OpenAgent Version

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.

OpenAgent integration: This skill integrates approval gates from OpenAgent's safety-first philosophy. You will request approval before each major phase.

Announce at start: "I'm using the openagent-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

Workflow with Approval Gates

dot

digraph writing_plans_openagent {
    rankdir=TB;
    start [label="Start:\nSpec/requirements", shape=ellipse];
    approval_explore [label="⏸️ REQUEST\nAPPROVAL\nto explore codebase", shape=diamond, style=filled, fillcolor="#ffffcc"];
    explore [label="Explore codebase\nIdentify files, patterns", shape=box];
    approval_write [label="⏸️ REQUEST\nAPPROVAL\nto write plan", shape=diamond, style=filled, fillcolor="#ffffcc"];
    write [label="Write plan\nBite-sized tasks", shape=box];
    approval_review [label="⏸️ REQUEST\nREVIEW\nof complete plan", shape=diamond, style=filled, fillcolor="#ffffcc"];
    revise [label="Revise plan\nbased on feedback", shape=box];
    approval_save [label="⏸️ REQUEST\nAPPROVAL\nto save plan file", shape=diamond, style=filled, fillcolor="#ffffcc"];
    save [label="Save plan to\ndocs/plans/", shape=box];
    approval_commit [label="⏸️ REQUEST\nAPPROVAL\nto commit plan", shape=diamond, style=filled, fillcolor="#ffffcc"];
    commit [label="Commit plan\nto git", shape=box];
    approval_execute [label="⏸️ OFFER\nexecution choice", shape=diamond, style=filled, fillcolor="#ffffcc"];
    done [label="Done", shape=ellipse];

    start -> approval_explore;
    approval_explore -> explore [label="approved"];
    explore -> approval_write;
    approval_write -> write [label="approved"];
    write -> approval_review;
    approval_review -> revise [label="changes\nrequested"];
    approval_review -> approval_save [label="approved"];
    revise -> approval_review;
    approval_save -> save [label="approved"];
    save -> approval_commit;
    approval_commit -> commit [label="approved"];
    commit -> approval_execute;
    approval_execute -> done;
}

Phase 1: Explore Codebase

⏸️ REQUEST APPROVAL: "May I explore the codebase to identify files, patterns, and architecture for this implementation?"

After approval:

•Identify relevant files and their purposes
•Find existing patterns to follow
•Locate test files and testing conventions
•Document tech stack and dependencies
•Note any existing similar features

Phase 2: Write Plan

⏸️ REQUEST APPROVAL: "May I write the implementation plan with bite-sized tasks following TDD?"

After approval, create plan following structure below.

Phase 3: Review Plan

⏸️ REQUEST REVIEW: "Plan complete. Here's the summary:

•[Number] tasks identified
•Files to create: [list]
•Files to modify: [list]
•Key architectural decisions: [summary]

May I show you the complete plan for review?"

After showing plan: Wait for approval or revision feedback.

Phase 4: Save Plan

⏸️ REQUEST APPROVAL: "May I save this plan to docs/plans/YYYY-MM-DD-<feature-name>.md?"

After approval: Save the plan file.

Phase 5: Commit Plan

⏸️ REQUEST APPROVAL: "May I commit the plan? Files: [list files]"

After approval:

bash

git add docs/plans/YYYY-MM-DD-<feature-name>.md
git commit -m "docs: add implementation plan for <feature-name>"

Phase 6: Offer Execution Choice

⏸️ OFFER EXECUTION CHOICE: (See Execution Handoff section below)

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:

markdown

# [Feature Name] Implementation Plan

> **For Claude:** REQUIRED SUB-SKILL: Use custom/openagent-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]

---

Task Structure

markdown

### 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**

```python
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

python

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

bash

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

code

## 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

## Execution Handoff

After committing the plan, offer execution choice:

**⏸️ 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 custom/openagent-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 custom/openagent-executing-plans

## OpenAgent Approval Gate Summary

**Approval points in planning workflow:**
1. ⏸️ Before exploring codebase
2. ⏸️ Before writing plan
3. ⏸️ After plan complete (review before saving)
4. ⏸️ Before saving plan file
5. ⏸️ Before committing plan
6. ⏸️ Before offering execution choice

**Why approval gates matter:**
- Prevents unnecessary codebase exploration
- Ensures user awareness of plan structure
- Creates natural checkpoints for review
- Aligns with OpenAgent's safety-first philosophy
- Allows user to guide planning direction

## Red Flags - Skipping Approval Gates

- Starting codebase exploration without approval
- Writing plan without approval
- Saving plan without review
- Committing without approval
- **"This is just planning, doesn't need approval"**
- **"I'll show it after it's complete"**
- **"Approval gates slow down planning"**
- **"Skipping gates to save time"**

**All of these mean: Stop. Request approval at each phase.**

## Final Rule

Planning with approval gates → user awareness at each phase → OpenAgent planning Otherwise → not OpenAgent planning

code


No exceptions without your human partner's permission.