lidge-jun

SVG diagrams, charts, and interactive visualizations for chat UI

cli-jaw Design workspace: create, preview, run, and export design pages from the right sidebar. Covers panel UX, direct-write workflow, artifact lifecycle, wireframe generation, design system, and Open Design adapter.

Unified desktop + browser automation. Routes DOM targets to CDP (cli-jaw browser), desktop apps to Computer Use, hybrid combos to both. Codex desktop/CLI required for Computer Use; macOS is app-scoped and Windows is window-scoped.

Structured browser web-ai workflow for ChatGPT, Gemini, and Grok in cli-jaw.

HWP/HWPX create, read, edit, review, template-fill, QA. Triggers: 한글, .hwp, .hwpx, HWP, HWPX, Korean documents, 한컴오피스, OWPML.

PDF RAG pipeline: text extraction + Apple Vision OCR + API embedding + cosine search. Triggers: "pdf 읽기", "pdf rag", "문서 검색", "pdf 분석", "pdf 청크", "pdf-expert", "pdf 임베딩"

Run autonomous multi-step research via Google Gemini Deep Research Agent. Produces cited reports from web sources.

Guide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine…

Ask AI web UIs (ChatGPT Chat and Work, Gemini, Grok) via standalone agbrowse browser automation. Chat model-family/tier selection, Work Power submission, reasoning-effort control, session resume, file/context upload, polling, copy-markdown fallback, and…

Chrome/CDP browser control and browser QA: open and smoke-test web UIs, inspect console/network state, fetch URLs, extract schema-bound data, snapshot, click, type, and screenshot. Prefer agbrowse for ad-hoc Playwright/browser QA before adding a separate…

Search discipline and proof engine for CLI agents: discover, prove, and deep-research with agbrowse. Verify original pages, score evidence, and optionally escalate to web-ai. Prefer agbrowse for public-web proof and URL QA before adding a Playwright runner or…

Vision-based coordinate click: screenshot → Codex CLI (NDJSON) → DPR correction → mouse click. Codex CLI only. Triggers: vision click, 비전 클릭, coordinate click, 좌표 클릭, screenshot click, non-DOM click, agbrowse-vision-click. NOT for: regular DOM clicks (use…

Find frontend UI effects and interaction patterns and return runnable code. Use when the user asks for a UI pattern or effect — e.g. "build a bottom sheet", "scroll reveal effect code", "mobile modal pattern", "sticky CTA bar", "hover card animation",…

Query and recommend design isms (styles). Use when the user asks about design styles, movements, or their concrete build tokens — e.g. "minimalism site", "brutalism color palette", "art deco font pairing", "which design style fits my project", "ism", "design…

Operate opencodex (`ocx`): discover CLI tasks offline, inspect local configuration, and manage a running proxy’s account pool, providers, models, routing, usage report, and management API. Use for proxy operation rather than editing the opencodex codebase.

Activate an xAI OAuth session as a local Grok proxy and CLI tool surface. Use when you need Grok models (chat, reasoning, search, images, video, voice, RAG) from OpenAI-compatible clients or agent tools. Requires a SuperGrok subscription + `progrok login`…

Use Aside's signed-in browser from Codex, Claude Code or another coding agent. Route authenticated browsing, explicit Aside code-mode batches and visual artifact work between aside-codemode, Aside REPL and Aside's agent. Use when the user asks for Aside or…

Query-first Windows failure intelligence: preflight the fuck-powershell landmine corpus before patches, and avoid PowerShell landmines when writing or running Windows shell commands: alias traps (curl/wget), POSIX redirects (/dev/null), native stderr…