junior-to-senior

Reviews agent-generated plans adversarially and rewrites them using codebase and web research.

Updated May 26, 2026
One-click install
npx skills add https://github.com/KinalVR-Expo2026/KinalVR-Client --skill junior-to-senior-kinalvr-expo2026
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: junior-to-senior
Source: https://github.com/KinalVR-Expo2026/KinalVR-Client/tree/main/.claude/.agents/skills/junior-to-senior
Command: npx skills add https://github.com/KinalVR-Expo2026/KinalVR-Client --skill junior-to-senior-kinalvr-expo2026

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Agent-generated plans often fail at two altitudes: they are either too vague ("add caching", "handle auth") or lost in low-level detail with no product vision. This Skill treats the plan as junior work, builds a senior reviewer grounded in live codebase research and current best practices, then diagnoses and rewrites the plan into a scoped, evidence-backed version. ## Core Features & Use Cases - Altitude Diagnosis: Classifies each plan section as fog (too vague), tunnel (too granular), or mixed using concrete checklists and a vague-word blacklist. - Evidence-Based Adversarial Review: Every finding cites a file and line from the repository or a dated web source, with blocker/major/minor severities and concrete fixes. - Plan Promotion: Produces a rewritten senior plan with goals, decisions with rationale, interfaces, sequencing, rollback strategy, and open questions for humans. - Use Case: After an agent drafts a migration plan, invoke the review to catch that the plan targets an API the pinned dependency version does not have, then receive a corrected, sequenced rewrite. ## Quick Start Ask the assistant to run a junior-to-senior senior review on the plan it just produced before committing to it.

Frequently Asked Questions about junior-to-senior

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

FAQPage Schema
How do I get a senior-level review of an AI-generated plan?

Invoke the junior-to-senior review on the plan the agent just produced. It freezes the artifact, researches the codebase and current best practices, then returns severity-classified findings and a rewritten senior version of the plan.

What makes a plan too vague or too detailed to execute?

A vague plan fails the start-tomorrow test: an engineer would still need to make architecture or product decisions themselves. An overly detailed plan lacks goals, success criteria, non-goals, and justification for the chosen approach over alternatives.

Does the review work without web access?

Yes. It runs full codebase research against lockfiles, conventions, and prior decisions, then marks every best-practice claim as training-data and unverified. It also names the decisions most likely to have shifted so you can spot-check them.

When should I not use adversarial plan review?

Skip it when there is no concrete artifact to review, since the process requires a frozen plan, design doc, or RFC as its target. It also does not invent product direction; genuine product choices are returned as open questions for a human.

How are review findings severity classified?

Findings are blockers when the plan fails as written, majors when it works but is worse than current state of the art or misfit to the repo, and minors for polish. Every finding carries evidence and a concrete fix.