Chachamaru127/claude-code-harness

Configures 2-Agent workflow between PM and implementation roles. Use when user mentions 2-Agent setup, PM coordination, Cursor setup, or 2-agent operations. Do NOT load for: solo operation, workflow execution, or handoff processing.

Implements authentication and payment features using Clerk, Supabase Auth, or Stripe. Use when user mentions login, authentication, payments, subscriptions, or Stripe. Do NOT load for: general UI work, database design, or non-auth features.

Diagnoses and fixes CI/CD pipeline failures. Use when user mentions CI failures, build errors, test failures, or pipeline issues. Do NOT load for: local builds, standard implementation work, reviews, or setup.

Integrates OpenAI Codex CLI as an MCP server to provide second-opinion reviews. Triggered by requests for Codex review, second opinion, or Codex setup. Do NOT load for: standard reviews without Codex, implementation work, or non-setup tasks.

Accesses the claude-mem MCP server from Cursor to search session history and record observations. Triggers: memory search, claude-mem, past decisions, record this. Do NOT load for: normal coding, temporary notes, or implementation work.

Sets up deployment, analytics, and health monitoring for projects. Use when user mentions deployment, Vercel, Netlify, analytics, or health checks. Do NOT load for: implementation work, local development, reviews, or setup.

Browser automation with persistent page state. Use when users ask to navigate websites, fill forms, take screenshots, extract web data, test web apps, or automate browser workflows. Trigger phrases include 'go to [url]', 'click on', 'fill out the form', 'take a screenshot', 'scrape', 'automate', 'test the website', 'log into', or any browser interaction request.

Generates documentation files including NotebookLM YAML and slide content. Use when user mentions documentation, YAML, NotebookLM, slides, or presentations. Do NOT load for: implementation work, code fixes, reviews, or deployments.

Implements features and writes code based on Plans.md tasks. Use when user mentions implementation, adding features, writing code, or creating new functions. Do not use for review or build verification.

Cleans up and organizes project files. Use when user mentions cleanup, archive, file bloat, or organizing Plans.md/session logs. Do NOT load for: implementation work, reviews, new feature development, or deployment.

Manages memory, SSOT files, and Plans.md operations. Use when user mentions memory, SSOT, decisions.md, patterns.md, merging, or migration. Do NOT load for: implementation work, reviews, ad-hoc notes, or in-session logging.

Provides development principles, guidelines, and VibeCoder guidance. Use when user mentions principles, guidelines, safety, or diff-aware editing. Do not use for actual implementation—use the impl skill instead.

Reviews code for quality, security, performance, and accessibility issues. Use when user mentions reviews, code review, security, performance, quality checks, PRs, diffs, or change review. Do NOT load for: implementation work, new feature development, bug fixes, or setup.

Controls session resume/fork for /work based on flags. Use only from workflow.

Manages session state transitions per SESSION_ORCHESTRATION.md. Internal use only.

Sets up new projects and generates workflow files like CLAUDE.md, AGENTS.md, Plans.md. Use when user mentions setup, initialization, new projects, or workflow file generation. Do NOT load for: implementation work, reviews, build verification, or deployments.

Guides diagnosis and resolution when problems occur. Use when user mentions something broken, errors, or that it doesn't work. Do NOT load for: successful builds, new feature implementation, or reviews.

Generates UI components and feedback forms. Use when user mentions components, UI, hero sections, forms, feedback, or contact requests. Do NOT load for: authentication features, backend implementation, database operations, or business logic.

Verifies builds, recovers from errors, and applies review fixes. Use when user mentions build verification, error recovery, applying review fixes, test failures, lint errors, or CI breaks. Do NOT load for: implementation work, reviews, setup, or new feature development.

Guides VibeCoder (non-technical users) through natural language development. Use when user asks what to do next, how to use the system, needs help, or is stuck. Do NOT load for: technical-user work, direct implementation requests, or reviews.

Manages workflow transitions including handoffs between PM and implementation roles, and auto-fixes review comments. Use when user mentions handoffs, reporting to PM, handing off to implementation, completion reports, or auto-fix. Do not use for 2-Agent setup—use 2agent skill instead.