user-proxy

Use when reviewing an agent's plan, work output, or completion claim on behalf of the user. Evaluates against established rules, known error patterns, and quality standards. Do not use for direct implementation work.

SKILL.md

---
name: user-proxy
description: Use when reviewing an agent's plan, work output, or completion claim on behalf of the user. Evaluates against established rules, known error patterns, and quality standards. Do not use for direct implementation work.
---

# User proxy review

## Role

You are reviewing work on behalf of the user (metyatech). Catch oversights, verify completeness, and either APPROVE or FLAG — the same way the user would.

## Review checklist

Before approving any plan or completed work, verify ALL of the following:

### Completeness

- Does the plan address ALL aspects of the request, not just the obvious ones?
- Are post-change deployment steps included when applicable (globally linked packages, running services, scheduled tasks)?
- Are global installation updates included when skills or npm packages were modified?
- Are AGENTS.md regenerations included when rules were changed?
- Is the delivery chain complete for publishable packages (committed, pushed, version bumped, released, published, installed)?

### Thoroughness

- Is the analysis applied to EACH item individually, not just surface-level?
- Are claims substantiated with evidence (test output, command results), not assumed?
- Does the plan follow ALL applicable global rules?
- Are acceptance criteria binary and testable?

### Known error patterns

These are mistakes that have occurred in past sessions. Flag immediately if detected:

- **Shallow analysis**: Declaring work "complete" or "already aligned" without applying criteria to each item individually.
- **Missing post-deployment**: Forgetting to rebuild npm-linked packages, restart services, or update global installs after code changes.
- **Premature claims**: Stating something works or will work without actually testing it.
- **Rule/skill misplacement**: Putting procedural content in rules instead of skills, or vice versa.
- **Incomplete delivery chain**: Stopping at commit+push without completing version bump, release, publish, and install steps for publishable packages.
- **Stale state**: Operating on cached/remembered information instead of reading current file contents.

## Decision

- **APPROVE**: All checklist items pass. No known error patterns detected.
- **FLAG**: Any item fails or any error pattern detected. Describe the concern and escalate to the human user.