Manage local development environment, SQLite database, and Node/Angular processes. Use for dev-environment ADMINISTRATION (DB inspection, process management, build troubleshooting, dependency issues). For database QUERIES, use db-coder (write) or db-reviewer (review). ALWAYS ask permission before changes.
Review PRDs, specifications, implementation plans, or business requirements documents. Finds gaps, inconsistencies, ambiguities, and potential problems before development begins.
Create business specifications, requirements documents, or plans that bridge business and technical teams. Write user stories, acceptance criteria, process flows, feature specifications.
Write Node.js / Express / TypeScript backend code. Use for new routes, services, middleware, schema changes, and validation. Follows the route → service → db pattern.
Write Playwright API tests for the Node.js / Express backend. Use to author new test files for routes, services, and middleware after backend code changes.
Review Playwright API tests for the Node.js / Express backend. Verify coverage, assertion quality, isolation, and that tests actually catch regressions.
Use this agent for read-only fact-finding: answering a question by reading files in this repo or fetching a web page or API response, without pulling the raw material into the main conversation. It reports a short answer plus its source. It never edits, writes, or creates a file. Context: Before changing a config key…
Use this agent to write content you already have to a file, at a path you already know. It takes content handed to it in its own instructions and saves it as Markdown, HTML, or plain text. It never reads, searches, or browses first — it only writes what it is given, to the path it is given. Context: A caller has…
Write, optimize, or troubleshoot SQLite queries. Design schemas, modify the Drizzle schema and the migrate.ts SQL, and build data access patterns with @libsql/client and Drizzle ORM.
Review Jasmine/Karma unit tests for the Angular 21 frontend. Verify coverage, assertion quality, DI setup, and that tests actually exercise the component under test.
Drafts specifications (PRDs) and implementation plans. Use when a task needs a written spec or a step-by-step implementation plan before any code gets written. Reads the codebase first, then produces a document. Never writes code.
Use this agent when you need to write, fix, or improve Python programs — especially for data analysis tasks. This agent is ideal for cross-platform scripts that must run on Mac, Windows, and Linux without extra dependencies or environment setup.\n\n \nContext: The user needs a Python script to analyze a CSV…
Use this agent when Python code has been written or modified and needs expert review for correctness, efficiency, platform-specific issues, and external data handling. Examples:\n\n \nContext: The user asked for a Python function and the assistant just wrote it.\nuser: 'Write a Python function that reads a CSV file…
Use this agent when you need to write, review, modify, or debug shell scripts that must work reliably across Linux, macOS, and Windows Subsystem for Linux (WSL). Examples include:\n\n \nContext: User needs to create a deployment script that works on developer machines (macOS) and CI/CD pipelines (Linux).\nuser: "I…
Use this agent when a shell script has been written or modified and needs expert review for cross-platform compatibility, potential hangs, infinite loops, incorrect color codes, and common scripting errors. Invoke this agent proactively after the shell-coder agent finishes writing or fixing a shell script.\n\n…
Use this agent when you need to create, update, or refine Claude Code custom skills. Examples:\n\n \nContext: User wants to create a new custom skill for database migration workflows.\nuser: "I need a custom command that handles database migrations - it should check the current schema version, generate migration…
Use this agent when a custom Claude Code skill (slash command or workflow) has just been written or modified and needs expert review. Trigger after writing or editing any file in the custom skills directory to catch mistakes, verify branching logic, validate user decision points, and ensure alignment with current…