Ask Why skill
A skill replacing Plan mode in Codex and Claude Code: it asks rounds of 3-5 questions and keeps a decision log.
- 117 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The skill creates files in the working folder and has no other side effects.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/ask-whyDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skill creates files in the working folder and has no other side effects.
This is third-party code. Review the repository files before installing.
What it does
Ask Why is for cases where an agent in Plan mode decides for the user and the plan keeps being rewritten. The skill puts question quality ahead of speed: each round asks 3-5 questions with priorities. P0 blocks the design and needs an answer at once, P1 changes the core plan, and P2 can be assumed and confirmed later. After the reply the skill updates `PLAN.md`, `DECISIONS.md` and `OPEN_QUESTIONS.md`, holding the current plan, confirmed choices and unresolved questions. For the first three rounds it challenges assumptions and looks for weak spots, converging from the fourth. When no blocking questions remain it writes `FINAL_PLAN.md`. Formats for files, questions and summaries live in `references/formats.md`. The decision stays with the user.
Who it is for. Product managers, researchers and developers planning complex work with fuzzy requirements.
Good fit when
- The task is vague and early choices strongly affect the result.
- Several reasonable directions exist and trade-offs must be weighed.
- The plan should come with a log of confirmed decisions and open questions.
- An agent, skill or workflow is being designed and every fork must be walked.
Not a fit when
- A simple factual answer is needed.
- It is a routine code fix or direct execution.
- The task has a complete spec and only needs implementing.
Example request
Use Ask Why to plan a mobile app launch: ask the first round of questions with P0, P1 and P2 priorities.Limitations
The skill suits tasks where the user is willing to answer questions; for short requests it explains that the mode is not needed and does the work directly. It creates `PLAN.md`, `DECISIONS.md`, `OPEN_QUESTIONS.md` and `FINAL_PLAN.md` in the working folder, so tidy extra files at the end. Only the `Ask_Why/` subdirectory is the skill, while the repository root is a project page. Interface metadata sits in `agents/openai.yaml`. Comparisons with Brainstorming and GStack in the README show examples chosen by the author. The last commit dates from July 2026.
How to disable. Delete the `Ask_Why/` folder from the agent skills directory. Plan files already created remain in the working folder and are deleted manually.
FAQ
What do P0, P1 and P2 mean?
P0 blocks the design and needs an answer at once. P1 changes the core plan and is confirmed soon. P2 can be assumed and confirmed later.
Which files does it create?
`PLAN.md`, `DECISIONS.md`, `OPEN_QUESTIONS.md` and finally `FINAL_PLAN.md`.
How does it differ from Plan mode?
Decisions stay with the user: the agent shows a set of related questions at once so trade-offs can be weighed and earlier answers revisited.
When is it unnecessary?
For simple factual questions, routine fixes, translation and polishing, and tasks with a complete spec.
Where are the format templates?
In `Ask_Why/references/formats.md`: formats for state files, questions, stage summaries and the final plan.
Discussion
No questions or reviews yet.
Related
Skillsgrill-me: plan interview skill
grill-me
A Claude Code skill: the agent asks one multiple-choice question at a time until every branch of a plan or design is resolved
A skill that cross-examines a plan before work starts: numbered questions in rounds with a recommended answer and no changes until you confirm
A CLAUDE.md and AGENTS.md file with seven principles: the agent frames the problem and scope before it writes code
Plugin that intercepts the agent plan and opens it in a browser: strike, replace, insert or comment, then send the feedback back.
A tutor for any subject: diagnosis, a study plan, project tasks and mastery checks, with the results saved to Markdown files
Third Brain Skills: a knowledge base and verification for agents
Third Brain Skills
21 skills build an Obsidian-style knowledge base with provenance, daily loops and verification before any done claim
README in short
The README explains the idea, shows repository structure, the working loop and comparisons with Brainstorming and GStack. Sections cover use cases and installation for different IDEs.