agentsclimarketplace

Plan

Skill YoniChechik/claude-code-config/skills/plan

Explore codebase and create structured implementation planFrom its SKILL.md

Install
npx -y skills add YoniChechik/claude-code-config --skill plan

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.2 KB, 692 tokens by cl100k_base, as published. Nobody here has run it

Restate the question in fully concrete terms, making every implicit detail explicit. Then answer.

Plan Mode

Explore the codebase and create a structured implementation plan.

HARD CONSTRAINT: ONE FEATURE = ONE PR

MUST plan every feature to ship as a single PR. Never split into multiple PRs, never propose phased/follow-up PRs, never suggest "PR 1 / PR 2" sequencing.

  • All tasks in the plan land together in one branch, one PR.
  • If scope feels too big for one PR, ask the user to narrow the scope — do not silently split it.
  • Codex critique pass must respect this too: reject any suggestion that breaks the work into multiple PRs.

Feature description from user input

"$ARGUMENTS"

Process

Use a subagent to carry out the following steps:

Step 1: Explore & Plan

  • Explore existing code patterns and architecture
  • Identify related files and components
  • Understand dependencies and integration points
  • Use web search to find relevant information and examples
  • Ask the user multiple questions throughout the process:
    • Unclear scope or boundaries
    • Multiple valid technical approaches
    • Breaking changes or migration needed
    • Concerns and tradeoffs
  • Decide yourself:
    • Implementation details, file/function names, code organization, other obvious choices

Step 2: Write Plan File

Determine feature name from branch: FEATURE_NAME=$(git rev-parse --abbrev-ref HEAD)

Create plan-$FEATURE_NAME.md in the current working directory (the feature clone directory) with this structure:

# Feature: [Feature Name]

## TLDR
[2 lines typical, max 5 for complex features - WHAT and WHY in plain language]

## Research and References
1 paragraph for simple feature, up to 5 paragraphs for complex features. Add research, references, links to similar implementations, relevant documentation. Include tradeoffs and how this relates to existing codebase patterns and architecture.

### Task 1: [Short Description]
**What:**
- Action 1
- Action 2

### Task 2: [Short Description]
**What:**
- Action 1
- Action 2
  • Tasks should be as independent as possible, with minimal dependencies between them.
  • Tasks should be actionable and specific, not vague or high-level.
  • No human in the loop. You can ask/search for relevant CLIs or MCPs.

Codex critique pass

After the plan file is written, run a second-opinion pass before terminating:

  • Critique: Invoke the /codex skill on the just-written plan-$FEATURE_NAME.md. Ask Codex to flag weak spots, missing considerations, risky assumptions, bad task sequencing, and unclear acceptance criteria.
  • Triage: Separate valid points from noise; note any items that are open questions for the user rather than direct fixes.
  • Propose: Present a short bulleted list of proposed plan changes. Use AskUserQuestion for concrete accept/reject choices; otherwise plain text.
  • Apply: Make surgical edits to the plan .md for accepted changes — do not rewrite wholesale.
  • Done: Tell the user the plan was critiqued by Codex and updated, and point them to the final file.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 326,764. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.