Skip to content

senseidukes/seihouse-skills

v1.0.0MIT

Official AI Agent Skills Suite for SEIHouse Productions.

browser-synth-builder

Build a complete, runnable, playable browser-based synthesizer ("VST-style synth in the browser") from requests like "make me a playable VST synth", "build a synthesizer with a working keyboard", "create a web instrument / playable piano / Web Audio synth", or any ask for a sound-producing music instrument page with a modern instrument design. Use when the deliverable must produce sound immediately in a browser preview without a DAW, plugins, or installation. Covers Web Audio implementation, onscreen keyboards with touch/mouse/computer-keyboard input, sound controls, and polished instrument UI. Not for native/DAW plugin development (VST3/AU), MIDI hardware integrations, or DAW host compatibility.

creative-ui-translator

Translate loosely described UI, animation, motion, material, atmosphere, and interaction ideas into precise Kimi-native creative-engineering implementation prompts. Use when the user describes how an interface should feel, move, react, reveal, transform, or behave physically and needs that idea converted into an actionable frontend graphics specification; when the user proposes or refines an interactive visual idea, references an existing visual experience, asks for the correct technical vocabulary, or wants a copy-paste prompt for implementing a UI effect.

plan-mode-gate

Enforces a mandatory plan mode before any non-trivial code change. The agent must complete four phases — Reconnaissance, Specification & Constraints, Blast Radius & Impact Analysis, and an Actionable Checklist — and present the plan for approval before editing any file. Use when the user asks to modify, refactor, extend, debug, or add features to an existing codebase; when a change touches multiple files or shared interfaces; when the user says "plan first", "don't code yet", "think before you change anything", "enter plan mode", or "analyze the impact"; or whenever skipping impact analysis risks breaking existing behavior. Not for greenfield scaffolding, one-line fixes, or pure questions about code.

seihouse-codebase-conventions

Orient coding agents inside SEIHouse repositories before implementation. Use when a task refers to existing SEIHouse systems, Library architecture, SEN, Reader Chamber, Living Codex, story generation, media storage, themes, cultivation, relics, profiles, persistence, synchronization, or shared components without defining how those systems fit together. Also use when adding, refactoring, or reviewing code that must follow repository structure, naming, state, data-flow, persistence, compatibility, and ownership conventions. This skill maps stable SEIHouse product concepts onto the repository's current implementation, prevents duplicate systems and architectural drift, and favors the smallest coherent change that preserves existing behavior.

seihouse-components-performance

Audit and optimize the performance, resilience, loading behavior, and device adaptability of a specified component while preserving its intended design and functionality.

seihouse-gemini-verification-loop

Makes Gemini slow down, verify implementation work in multiple distinct passes, fix confirmed in-scope defects, and avoid unrequested additions or broad refactors.

seihouse-layout-optimization

Audit and optimize a specified component so its composition, hierarchy, usability, and visual quality remain intentional across mobile, tablet, laptop, desktop, portrait, landscape, narrow, wide, and short-height viewports.

seihouse-pr

Check branch scope, prepare completed repository changes, and create or update a pull request.

seihouse-pr-review

Perform a focused, token-efficient self-review immediately before opening or updating a pull request. Intended for implementation agents such as Codex, Claude, Kimi/Cami, and similar coding agents.

sensei-skill

Communicate technical work to SENSEI as a founder and product owner who is not a coder. Use during coding, debugging, implementation, reviews, pull requests, technical planning, and change summaries to explain outcomes in natural speech, surface meaningful product consequences or chain reactions, and recommend the most useful next step.

workshop-replica

Create a faithful, isolated Workshop replica of a real application page, component, or flow using local mock state while preserving portability back to the source app.