plan-interrogate

Stress-test implementation plans by resolving every open decision one question at a time.

2.8k|277|Updated Feb 1, 2026
One-click install
npx skills add https://github.com/rohitg00/pro-workflow --skill plan-interrogate
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: plan-interrogate
Source: https://github.com/rohitg00/pro-workflow/tree/main/skills/plan-interrogate
Command: npx skills add https://github.com/rohitg00/pro-workflow --skill plan-interrogate

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Plans often contain hidden undecided branches that surface mid-implementation, causing rework and stalled progress. This Skill walks a plan's decision tree node by node, forcing every open decision to be resolved, inferred from the codebase, or explicitly deferred before any code is written.

Core Features & Use Cases

  • Decision tree extraction: Converts a plan into a tree of branch points, each marked open or resolved with a source tag (user or inferred from the codebase).
  • One-question-at-a-time interrogation: Asks a single tight question per ready node, always paired with a recommended answer and reasoning.
  • Durable artifacts: Produces a decision ledger, an optional CONTEXT.md glossary of project terms, and numbered decision records in docs/decisions/ for contested choices.
  • Use Case: Before starting a multi-service refactor, run the interrogation to pin down database strategy, API versioning, and rollout order, then paste the resulting decision ledger directly into the plan document.

Quick Start

Ask the agent to interrogate your draft plan and resolve every open decision before implementation begins.

Frequently Asked Questions about plan-interrogate

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I stress-test a design plan before writing code?

Run plan-interrogate on the plan. It restates the plan, extracts every branch point into a decision tree, then asks one focused question per open node with a recommended answer until zero nodes remain open.

What is a decision ledger in software planning?

A decision ledger is a flat list of resolved decisions in the format Decision - Choice - Source (user or inferred: path). This Skill produces it as the primary output so you can paste it directly into the plan document.

Can the skill answer questions from the codebase instead of asking me?

Yes. Before asking, it checks whether the answer already exists in the codebase, prior commits, or existing docs. If found, the node is marked resolved with source inferred and the file path, skipping the question entirely.

What happens if I decline to make a decision during interrogation?

The node is marked DEFERRED with the reason you gave, which is distinct from open. The interrogation can still exit when no open nodes remain, and deferred items stay visible in the final ledger.

When should I write a decision record instead of just a ledger entry?

Write a decision record only for contested or hard-to-reverse choices a future reader would question. Records live in docs/decisions/NNNN-slug.md, numbered from the highest existing file, and trivial choices are skipped as noise.