plan-interrogate — Claude Code skill: what it does, how to install
The plan-interrogate skill probes a plan before implementation by mapping its decisions and asking about them one at a time. Use it to clarify unresolved design choices.
MIT ★ 2,966 updated 2026-09-29
/plugin marketplace add rohitg00/pro-workflow
/plugin install pro-workflow@pro-workflowWhat plan-interrogate does
The skill first restates the plan and asks the user to confirm that it has the right version. It then maps decisions and their dependencies, checks the codebase, commits, and existing docs for answers, and asks about one unresolved decision at a time. Each question comes with a suggested choice and a reason. Decisions settled by project evidence are marked with the path that supports them.
The final output is a flat decision ledger showing each choice and whether it came from the user or project evidence. The skill can also offer to capture project-specific terms in CONTEXT.md and contested or hard-to-reverse choices in docs/decisions/NNNN-slug.md. Before adding a decision record, it reads that directory and continues from the highest existing number. The skill guides planning; it does not implement the plan.
What it helps with
- surface unresolved choices before coding begins
- ask about decisions in dependency order
- look for answers in code, commits, and existing docs
- record each choice and its source
- capture project terminology and difficult decisions
Example prompts for the agent
- Pressure-test this plan, one decision at a time, and confirm your restatement before continuing.
- Map the decision dependencies, then recommend an option for each open choice.
- Finish with a decision ledger and offer to capture the project terms in
CONTEXT.md.
How to install plan-interrogate
1Plugin marketplace
/plugin marketplace add rohitg00/pro-workflow/plugin install pro-workflow@pro-workflowThis installs the whole pro-workflow plugin with all its skills
2Skills CLI (Vercel)
npx skills add rohitg00/pro-workflow --skill plan-interrogate -a claude-code3Manually: put the files in place
~/.claude/skills/plan-interrogate/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/plan-interrogate/
1Skills CLI (Vercel)
npx skills add rohitg00/pro-workflow --skill plan-interrogate -a codex2Manually: put the files in place
.agents/skills/plan-interrogate/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add rohitg00/pro-workflow --skill plan-interrogate -a cursor2Manually: put the files in place
.cursor/skills/plan-interrogate/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install rohitg00/pro-workflow skills/plan-interrogate2Manually: put the files in place
.github/skills/plan-interrogate/SKILL.md1Skills CLI (Vercel)
npx skills add rohitg00/pro-workflow --skill plan-interrogate -a gemini-cli2Platform CLI
gemini skills install https://github.com/rohitg00/pro-workflow --path skills/plan-interrogateManually: put the files in place
.devin/skills/plan-interrogate/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about plan-interrogate
How do I install plan-interrogate in Claude Code?
Run these commands in order: /plugin marketplace add rohitg00/pro-workflow, then /plugin install pro-workflow@pro-workflow.
What does plan-interrogate do?
It breaks a plan into decisions, checks project files for existing answers, and asks one question at a time with a recommendation and rationale. It outputs a ledger once every decision is resolved or deferred.
When does plan-interrogate write decision records?
It offers records for contested or hard-to-reverse decisions, using docs/decisions/NNNN-slug.md. It checks the directory first and continues numbering from the highest existing record.
Source text
SKILL.md · MIT
--- name: plan-interrogate description: Stress-test a plan by walking its decision tree one question at a time. Use when the user wants to pressure-test a design before implementation. user-invocable: true --- # plan-interrogate Drive a plan from sketch to commitment by resolving every open decision before any code is written. ## Method 1. Restate the plan in one paragraph. Confirm with the user that this is the plan being interrogated. Do not proceed on a mis-restatement. 2. Extract the decision tree. Every branch point becomes a node. Mark each node as **open** (undecided) or **resolved**. A resolved node carries a source tag: `user` (the user answered), `inferred` (the codebase or an existing constraint settled it). 3. Resolve in dependency order. A node is ready when every node it depends on is resolved. 4. For each ready open node, ask exactly one question. Keep the question tight and binary or small-multiple-choice when possible. 5. Pair every question with a **recommended answer** and one sentence of reasoning. The user can confirm, pick a different option, or push back. 6. Before asking, check whether the answer already lives in the codebase, prior commits, or an existing doc. If so, skip the question and mark the node resolved with source `inferred: <path>`. 7. Exit only when zero nodes are open. Print the resolved tree as a flat list: `Decision - Choice - Source (user | inferred: <path>)`. ## Anti-patterns - Asking multiple questions at once. The user loses context and you lose the ability to react to each answer individually. - Asking before exploring. If a fifteen-second read would answer the question, read first. - Asking without a recommendation. A question without a stance is a survey; it offloads design onto the user. - Rolling past an unresolved node. If a dependency is not pinned, the downstream question is premature. ## Outputs The interrogation produces three artifacts, not just answers. Offer to write each; do not force it. 1. **Decision ledger** (always). The resolved tree as a flat list: `Decision - Choice - Source (user | inferred: <path>)`. 2. **`CONTEXT.md`** (when the interrogation surfaced project-specific terms). A short shared-language file: every domain term you and the user had to pin down, with a one-line definition in the project's own words. This is what stops the agent from using twenty words where one will do next session, and keeps names in code consistent. One term per line: `term - what it means here`. Point future sessions at it. On re-run, merge new terms in place rather than overwriting existing ones. 3. **Decision records** (for contested or hard-to-reverse nodes only). One short record per decision that a future reader would question: the context, the choice, the alternatives rejected, and why. Keep them in `docs/decisions/NNNN-slug.md`. Read the directory first and number from the highest existing record so two records never collide. Skip the obvious ones - a record for a trivial choice is noise. ## Output contract The decision ledger the user can paste into the plan doc. No prose summary. No hedging. If the user declines to decide a node, mark it `DEFERRED` with the reason the user gave - this is not the same as open. When you write `CONTEXT.md` or a decision record, keep it in the project's language, not a generic template.
© Rohit Ghumare · MIT · rohitg00/pro-workflow