pde-review-companion

Review PDE plan artifacts for missed intent and misinterpreted requirements.

1|2|Updated Apr 13, 2026
One-click install
npx skills add https://github.com/jgwill/dotagents --skill pde-review-companion
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pde-review-companion
Source: https://github.com/jgwill/dotagents/tree/main/skills/pde-review-companion
Command: npx skills add https://github.com/jgwill/dotagents --skill pde-review-companion

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents wasted effort and missed requirements by helping a human review freshly generated PDE outputs before the sleeping agent begins executing on them.

Core Features & Use Cases

  • Missed-intent detection: Compares PDE “Secondary Intents” and ambiguity flags against the original project input to spot gaps, over-specific interpretation, and incorrect repo/file boundaries.
  • Actionable edit suggestions: Recommends concrete changes to the ordering and dependencies inside .pde/*.md to improve the plan the agent will follow.
  • Review-window coordination: Helps the human stay within the limited wake window (e.g., 12 minutes) and optionally advises when to manually wake the agent.

Quick Start

Use the pde-review-companion skill with your PDE output in .pde/ to audit intent coverage and propose targeted edits to the .pde/*.md plan.

Frequently Asked Questions about pde-review-companion

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

FAQPage Schema
How do I review agent generated plans to catch missed intent before execution?

To catch missed intent, compare the PDE plan artifacts in `.pde/*.md` and `.pde/*.json` against the original root input file to identify gaps, incorrect boundaries, and over-specific interpretations before the agent begins execution.

What is a PDE review window for multi-agent orchestration?

A PDE review window is a limited wake period, such as 12 minutes, occurring when agents have decomposed work and are sleeping, allowing humans to audit intent coverage and suggest edits to the plan before execution resumes.

How to validate secondary intents in markdown plan files against original requirements?

Validating secondary intents requires comparing the PDE “Secondary Intents” and ambiguity flags in `.pde/*.md` files against the original project input to spot gaps, over-specific interpretation, and incorrect repo or file boundaries.

Can I edit `.pde/*.md` plan ordering and dependencies during the agent sleep window?

Yes, you can edit `.pde/*.md` plan ordering and dependencies during the agent sleep window, using actionable edit suggestions to improve the plan while respecting the wake timer and optionally advising when to manually wake the agent.

What are the limitations of automated plan review for multi-agent repo workflows?

Automated plan review is limited by the execution wake timer, requires reading specific PDE artifact formats, and depends on comparing outputs to the original root input file to produce human-ready feedback rather than executing changes directly.