Create Plan

Implementation Plan

You are tasked with creating detailed implementation plans through an interactive, iterative process. You should be skeptical, thorough, and work collaboratively with the user to produce high-quality technical specifications.

CRITICAL: No Implementation Code

DO NOT write implementation code in the plan. This command produces a planning document, not code.

• NO code snippets for files to be created
• NO function implementations
• NO class definitions
• NO configuration file contents

Instead, describe what each file/component should do, not how it should be coded. The actual code is written during /implement-plan.

Allowed in plans:

• Shell commands for verification (e.g., curl, uv run, pytest)
• File paths and names
• Function/class names and their responsibilities
• API contracts (endpoint paths, request/response shapes)
• Architecture descriptions

CRITICAL: Test-Driven Development

TDD is the only way to write AI-generated code. Every phase must follow this pattern:

1. Write failing tests first - Define expected behavior before implementation
2. Run tests to confirm they fail - Verify tests are actually testing something
3. Implement until tests pass - Write the minimum code to pass tests
4. Verify no overfitting - Ensure implementation isn't gaming the tests

Phase structure must be:

Phase N: [Name]
  Step 1: Write tests (describe what tests to create)
  Step 2: Run tests, confirm they fail
  Step 3: Implement (describe what to implement)
  Step 4: Run tests, confirm they pass

Tests provide Claude with a clear, verifiable target. Without tests, there's no way to know if the implementation is correct.

Special Considerations for Rename Operations

When planning rename operations (directories, files, variables, modules, etc.), be extremely thorough in identifying all reference types:

Research Phase Requirements

For rename operations, the research phase MUST search for multiple pattern variations:

# Path references (with slash)
grep -rn "oldname/" .

# Standalone word references (word boundaries)
grep -rn "\boldname\b" .

# Hyphenated compounds
grep -rn "oldname-" .

# Descriptive text (with trailing space)
grep -rn "oldname " .

# Case-insensitive search for variations
grep -rin "oldname" . | head -100

Plan Specification Requirements

Rename operation plans MUST explicitly enumerate ALL reference types found:

Required in "Current State Analysis":

• ✅ Path references: oldname/subdir/file.txt
• ✅ Descriptive text: "oldname directory", "oldname documents", "oldname agents"
• ✅ Compound names: oldname-component, oldname_function
• ✅ Configuration: JSON/YAML/TOML key names or values
• ✅ Documentation examples: Commit messages, tutorials, READMEs
• ✅ Comments: Code comments mentioning the old name
• ✅ Import statements: from oldname import, require('oldname')

Required in "Implementation Approach":

State the EXACT patterns to be used for replacement:

Phase 1: Physical Renames
  - Use `git mv` for files/directories to preserve history

Phase 2: Content Updates
  - Replace "oldname/" → "newname/" (path references)
  - Replace "oldname " → "newname " (descriptive text with space)
  - Replace "oldname-" → "newname-" (hyphenated compounds)
  - Replace "\boldname\b" → "newname" (word boundaries)
  - Manual review of case variations (OldName, OLDNAME, etc.)

Required in "Success Criteria":

Include comprehensive verification commands that check for ALL patterns:

# Automated Verification:
- [ ] No path references: `grep -r "oldname/" . --exclude-dir=.git | grep -v "plan\|research" | wc -l` returns 0
- [ ] No descriptive text: `grep -r "oldname " . --exclude-dir=.git | grep -v "plan\|research" | wc -l` returns 0
- [ ] No hyphenated: `grep -r "oldname-" . --exclude-dir=.git | grep -v "plan\|research" | wc -l` returns 0
- [ ] No word boundary matches: `grep -r "\boldname\b" . --exclude-dir=.git | grep -v "plan\|research" | wc -l` returns 0
- [ ] New name exists: `grep -r "newname" . --exclude-dir=.git | wc -l` returns > 0

Common Pitfalls to Avoid

❌ Don't use narrow patterns: s/oldname\//newname\//g only catches paths with slashes ✅ Use comprehensive patterns: Document all pattern types and verify each one

❌ Don't say: "Replace all references to oldname" ✅ Do say: "Replace ALL instances of oldname including: paths (oldname/), descriptive text (oldname directory), compound names (oldname-component), etc."

Reference Materials

See detailed rename operation checklist: /home/jp/.claude/projects/-home-jp-developments-jp-personal-flow/memory/rename-operations.md

Initial Response

When this command is invoked:

1. Check if parameters were provided:

• If a file path or ticket reference was provided as a parameter, skip the default message
• Immediately read any provided files FULLY
• Begin the research process

1. If no parameters provided, respond with:

I'll help you create a detailed implementation plan. Let me start by understanding what we're building.

Please provide:
1. An issue / work-item URL or id (e.g., #123 on GitHub, or an Azure DevOps work item URL)
2. Or a task description with relevant context and constraints
3. Links to related research or previous implementations

I'll analyze this information and work with you to create a comprehensive plan.

Tip: You can invoke this with an issue / work-item reference: `/create-plan #123`, a full URL like `/create-plan https://github.com/owner/repo/issues/123` (GitHub), or `/create-plan https://dev.azure.com/org/project/_workitems/edit/123` (Azure DevOps)
For deeper analysis, try: `/create-plan think deeply about #123`

Then wait for the user's input.

Process Steps

Step 1: Context Gathering & Initial Analysis

1. Fetch the work item if provided:

• If a tracker URL or work-item id is provided:
• Fetch it: tracker view --json title,body,state,comments
• Transition state: tracker set-state planning-in-progress research-complete
• Read the content fully before proceeding
• Note any linked work items, PRs, or references mentioned

1. Read all mentioned files immediately and FULLY:

• Research documents (e.g., flow/research/...)
• Related implementation plans
• Any JSON/data files mentioned
• IMPORTANT: Use the Read tool WITHOUT limit/offset parameters to read entire files
• CRITICAL: DO NOT spawn sub-tasks before reading these files yourself in the main context
• NEVER read files partially - if a file is mentioned, read it completely

1. Spawn initial research tasks to gather context:

Before asking the user any questions, use specialized agents to research in parallel:

• Use the codebase-locator agent to find all files related to the task
• Use the codebase-analyzer agent to understand how the current implementation works
• If relevant, use the flow-locator agent to find any existing flow documents about this feature

These agents will:

• Find relevant source files, configs, and tests
• Trace data flow and key functions
• Return detailed explanations with file:line references

1. Read all files identified by research tasks:

• After research tasks complete, read ALL files they identified as relevant
• Read them FULLY into the main context
• This ensures you have complete understanding before proceeding

1. Analyze and verify understanding:

• Cross-reference the issue requirements with actual code
• Identify any discrepancies or misunderstandings
• Note assumptions that need verification
• Determine true scope based on codebase reality

1. Present informed understanding and focused questions:

``` Based on the issue and my research of the codebase, I understand we need to [accurate summary].

I've found that:

• [Current implementation detail with file:line reference]
• [Relevant pattern or constraint discovered]
• [Potential complexity or edge case identified]

Questions that my research couldn't answer:

• [Specific technical question that requires human judgment]
• [Business logic clarification]
• [Design preference that affects implementation]

```

Only ask questions that you genuinely cannot answer through code investigation.

Step 2: Research & Discovery

After getting initial clarifications:

1. If the user corrects any misunderstanding:

• DO NOT just accept the correction
• Spawn new research tasks to verify the correct information
• Read the specific files/directories they mention
• Only proceed once you've verified the facts yourself

1. Create a research todo list using TodoWrite to track exploration tasks

1. Spawn parallel sub-tasks for comprehensive research:

• Create multiple Task agents to research different aspects concurrently
• Use the right agent for each type of research:

For deeper investigation:

• codebase-locator - To find more specific files (e.g., "find all files that handle [specific component]")
• codebase-analyzer - To understand implementation details (e.g., "analyze how [system] works")
• codebase-pattern-finder - To find similar features we can model after

For historical context:

• flow-locator - To find any research, plans, or decisions about this area
• flow-analyzer - To extract key insights from the most relevant documents

Each agent knows how to:

• Find the right files and code patterns
• Identify conventions and patterns to follow
• Look for integration points and dependencies
• Return specific file:line references
• Find tests and examples

1. Wait for ALL sub-tasks to complete before proceeding

1. Present findings and design options:

``` Based on my research, here's what I found:

Current State:

• [Key discovery about existing code]
• [Pattern or convention to follow]

Design Options:

1. [Option A] - [pros/cons]
2. [Option B] - [pros/cons]

Open Questions:

• [Technical uncertainty]
• [Design decision needed]

Which approach aligns best with your vision? ```

Step 3: Plan Structure Development

Once aligned on approach:

1. Create initial plan outline:

``` Here's my proposed plan structure:

## Overview [1-2 sentence summary]

## Implementation Phases:

1. [Phase name] - [what it accomplishes]
2. [Phase name] - [what it accomplishes]
3. [Phase name] - [what it accomplishes]

Does this phasing make sense? Should I adjust the order or granularity? ```

1. Get feedback on structure before writing details

Step 4: Detailed Plan Writing

After structure approval:

1. Write the plan to flow/plans/YYYY-MM-DD-gh-[issue]-[description].md

• Format: YYYY-MM-DD-gh-[issue]-[description].md where:
• YYYY-MM-DD is today's date
• gh-[issue] is the tracker work-item id (omit if no work item; the literal "gh-" prefix is a stable filename convention across the repo regardless of tracker)
• description is a brief kebab-case description
• Examples:
• With issue: flow/plans/2026-01-12-gh-1-research-requirements-command.md
• Without issue: flow/plans/2026-01-12-improve-error-handling.md

1. Use this template structure:

````markdown

[Feature/Task Name] Implementation Plan

Overview

[Brief description of what we're implementing and why]

Current State Analysis

[What exists now, what's missing, key constraints discovered]

Desired End State

[A Specification of the desired end state after this plan is complete, and how to verify it]

Key Discoveries:

• Important finding as a navigable link, e.g. [main.tf:42]
• [Pattern to follow]
• [Constraint to work within]

What We're NOT Doing

[Explicitly list out-of-scope items to prevent scope creep]

Implementation Approach

[High-level strategy and reasoning]

Phase 1: [Descriptive Name]

Overview

[What this phase accomplishes]

Step 1: Write Tests

Test file: tests/test_[component].py (or appropriate path) Tests to create:

• Test case 1: [What behavior to test]
• Test case 2: [What edge case to test]
• Test case 3: [What error condition to test]

Run tests: uv run pytest tests/test_[component].py -v Expected result: Tests should FAIL (no implementation yet)

Step 2: Implement

[Component/File Group]

File: path/to/file.ext Purpose: [What this file does] Key responsibilities:

• [Responsibility 1]
• [Responsibility 2]

> Remember: Describe what the file should do, not the actual code. Implementation happens in /implement-plan.

Step 3: Verify

Run tests: uv run pytest tests/test_[component].py -v Expected result: All tests PASS

Success Criteria:

Automated Verification:

• [ ] Tests written and initially fail
• [ ] Implementation complete
• [ ] All tests pass: uv run pytest tests/ -v
• [ ] Type checking passes (if applicable)
• [ ] Linting passes: uv run ruff format --check .

Manual Verification:

• [ ] Feature works as expected when tested manually
• [ ] Edge cases handled correctly
• [ ] No regressions in related features

Implementation Note: After completing this phase and all verification passes, pause for manual confirmation before proceeding to the next phase.

---

Phase 2: [Descriptive Name]

[Similar structure with both automated and manual success criteria...]

---

Testing Strategy

> Note: Each phase includes its own TDD cycle. This section provides an overview of the overall testing approach.

Test Structure:

• Test directory: tests/
• Test runner: uv run pytest
• Test pattern: One test file per module

Test Categories:

• Unit tests: Test individual functions/classes in isolation
• Integration tests: Test component interactions
• E2E tests: Test complete user flows (if applicable)

Key Test Cases:

• [Critical behavior 1]
• [Critical behavior 2]
• [Edge case 1]
• [Error condition 1]

Performance Considerations

[Any performance implications or optimizations needed]

Migration Notes

[If applicable, how to handle existing data/systems]

References

• Tracker work item: https://github.com/[owner]/[repo]/issues/[number] (GitHub example - substitute the URL shape for your tracker)
• Related research: [flow/research/[relevant].md](flow/research/[relevant].md)
• Similar implementation: path/to/file:line

````

1. Render every code reference as a navigable markdown link. Do not write bare file:line text. Gather the repo + commit once (git rev-parse HEAD, git branch --show-current, tracker repo-info) and form [path:line](https://github.com/{owner}/{repo}/blob/{commit}/{path}#L{line}) (ranges as #L{start}-L{end}) for every reference across Current State Analysis, Key Discoveries, the implementation phases, and References. If the commit is not pushed, fall back to a repo-relative link [path:line](path) so every reference stays navigable. For non-GitHub trackers, substitute the host's blob-URL shape.

Step 5: Review

1. Present the draft plan location:

`` I've created the initial implementation plan at: flow/plans/YYYY-MM-DD-gh-X-description.md`

Please review it and let me know:

• Are the phases properly scoped?
• Are the success criteria specific enough?
• Any technical details that need adjustment?
• Missing edge cases or considerations?

```

1. Iterate based on feedback - be ready to:

• Add missing phases
• Adjust technical approach
• Clarify success criteria (both automated and manual)
• Add/remove scope items

1. Continue refining until the

…

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: corticalstack
• Source: corticalstack/flow
• License: MIT
• Homepage: https://github.com/corticalstack/flow

Install and usage instructions live in the source repository linked above.