amplify

Amplify

Skills and configuration for AI coding agents. Works with Claude Code, OpenCode, Codex, and Amp.

What's Inside

Skills

Skill	What It Does
ask-questions-if-underspecified	Requirement clarification. Ask focused questions before implementing.
browse-site	Browser automation using agent-browser CLI. Navigate, click, fill forms, screenshot, scrape.
code-simplifier	Simplifies and refines code for clarity, consistency, and maintainability while preserving behavior.
create-plan	Generate concise implementation plans. Scope, action items, open questions, documentation links.
create-spec	Create or update SPEC.md from requirements, notes, or interview output. Structures information into consistent spec format.
create-vault	Create Ampi-ready SQLite vaults from document folders or existing SQLite tables with search contracts.
execute-plan	Work through PLAN.md one task at a time. Questions before action, summaries after, learning persistence.
fetch-github-file	Fetch source code from GitHub URLs. Auto-triggers on GitHub file links and supports line ranges, directories, and full repo clones via gh CLI.
frontend-design	Build distinctive, production-grade interfaces that don't look like AI slop. Opinionated about typography, color, motion, and spatial composition.
frontend-slides	Create stunning, animation-rich HTML presentations, including PPT/PPTX-to-web workflows.
guide	Interactive teaching mode. Orchestrates clarification, planning, and guided execution.
interview	Deep requirements gathering through exhaustive questioning. Hands off to create-spec or create-plan when complete.
liveview-forms	Build correct, resilient forms in Phoenix LiveView. Validation, error feedback, recovery, nested forms, uploads.
liveview-optimistic-ui	Build LiveView interactions that feel instant. JS commands, streams, error recovery, race conditions, accessibility.
mcp-builder	Build high-quality MCP servers for Python (FastMCP) and Node/TypeScript MCP SDK integrations.
project-memory	Maintain durable, project-specific memory in MEMORY.md. Persists preferences, conventions, and lessons across sessions.
roadmap	Create a ROADMAP.md for open source projects. Vision, milestones, timeline (Now/Next/Later/Future). Follows open source conventions.
skill-creator	Meta-skill for building new skills. Structure, progressive disclosure, bundled resources.
smart-commit	Analyze unstaged changes, group into atomic commits, generate conventional commit messages.
spec-story	Transform a dense SPEC.md into an entertaining narrative. Same information, actually fun to read.
stop-slop	Remove AI writing patterns from prose. Eliminate filler phrases, formulaic structures, and predictable AI tells.
tech-docs-writer	Technical writer that transforms codebases into clear documentation. README files, API docs, architecture guides.
update-deps	Update Elixir Mix/Hex dependencies safely, including verification and migration handling.
write-test	Write meaningful tests using coverage as a guide. Prioritizes real user workflows over implementation details.

Agents

Agent	What It Does
create-plan	Create implementation plans through interviewing. Thinks deeply before planning.
execute-plan	Execute PLAN.md tasks one at a time with human oversight.
plan-reviewer	Deep analysis of PLAN.md before execution. Validates against codebase, checks task decomposition, identifies risks.

Scripts

Script	What It Does
wiggum	Autonomous coding loop (Ralph Wiggum technique). Iterates through a plan until complete.

Skill Workflows

Guided Learning: guide → ask-questions → create-plan

The guide skill orchestrates a full learning workflow: