chenxin-yan

Use for "how does X work", code walkthroughs before changing something, and placement / ownership / layering questions ("where should this live", "which package owns this", "is this the right layer"). Explains subsystem architecture, runtime flow, onboarding…

Map a codebase or run a resumable, dependency-first guided tour.

Critically evaluate GitHub PR review comments instead of blindly applying them. verify each against the code, fix what's valid, and report per-comment verdicts. Use when the user wants to handle, triage, or address PR review feedback.

Ground facts about a third-party dependency by reading its actual source at the installed version, instead of recalling it from memory. Clones the upstream repo into .agent-sources/ (git-excluded), checks out the tag matching the installed version, then…

Stage and commit changes with a Conventional Commits message. Matches the repo's existing style, writes a structured subject and body, and commits without push/force/amend. Use whenever the user wants to commit, ship, save progress, or close out a task —…

Render a plan, review, design, or any structured topic as a single skimmable HTML file in the current directory, built for visual review (diagrams, charts, stats, side-by-side, tables, callouts) instead of prose. Use whenever the user says "make an html for…

Manage demo deployments.

Build a sales funnel

Create and manage prd.json task lists for Ralph AI coding agent. Use when asked to "create tasks", "add task", "create PRD", "break down features", "plan implementation", or when managing Ralph project tasks.

Create or refine SPEC.md technical specifications for Ralph AI coding agent. Use when asked to "create spec", "write specification", "define requirements", "plan architecture", or when setting up a new project.