ce

Audit or bootstrap Claude Code configuration for a repository

Expert at code review for pull requests from technical, product, and DX perspectives. Use this agent when the user has completed work on a feature branch and needs review before merging. Analyzes all changes between branches, evaluates user impact, assesses developer experience, enforces project standards, and returns a flat list of fixes with no optional tier.

Writes articles, essays, and long-form content using professional journalistic frameworks. Takes a brief or topic, recommends the best story structure, asks clarifying questions, then drafts the piece section by section. Use when writing articles, blog posts, features, essays, opinion pieces, trend analyses, or any narrative prose that needs professional structure.

Use this agent when you want someone to poke holes in a plan, design, or idea before committing. Examples: <example> Context: User has a plan they want validated user: "Here's my plan for the new caching layer. What's wrong with it?" assistant: "I'll look for real flaws, risks, and unstated assumptions." <commentary> User wants rigorous critique. This agent looks harder for problems than a typical reviewer, but only raises genuine issues. </commentary> </example> <example> Context: User is considering a major decision user: "We're thinking of switching to DuckDB. Play devil's advocate." assistant: "I'll examine this critically and surface any real concerns." <commentary> User wants thorough scrutiny before committing. The goal is catching genuine issues, not arguing against the decision. </commentary> </example> <example> Context: User wants to find gaps in a spec user: "Review this spec and tell me what I'm missing" assistant: "I'll identify gaps, edge cases, and unstated assumptions." <commentary> Finding what's missing requires assuming something IS missing and looking for it. </commentary> </example>

Specialist at efficiently reading and analyzing large log files using targeted search and filtering. Optimized to avoid loading entire logs into context by using grep-style workflows, time and severity filters, and iterative refinement across arbitrary log formats.

Guides clean, scalable system architecture during the build phase. Use when designing modules, defining boundaries, structuring projects, managing dependencies, or preventing tight coupling and brittleness as systems grow.

Enforces precise, minimal design for dashboards and admin interfaces. Use when building SaaS UIs, data-heavy interfaces, or any product needing Jony Ive-level craft.

Executes implementation plans with smart task grouping. Groups related tasks to share context, parallelizes across independent subsystems.

Diagnose and fix tests that pass in isolation but fail when run concurrently. Covers shared state isolation, resource conflicts, and timing-based flakiness.

Prevents silent failures and context loss in error handling. Use when writing try-catch blocks, designing error propagation, reviewing catch blocks, or implementing Result patterns.