Astack — Habilidades Claude Code Curadas
ai-native-engineer/astack
Skills curadas para Claude Code: sessões, voice memos, iMessage, DevTools, Meta Ads, crawl, OSS
Ver no GitHubConjunto curado de skills para Claude Code, instalável como plugin. Inclui session history, voice memos, iMessage, Chrome DevTools, Meta Ads, web crawling, exploração OSS e coleta de contexto de projetos — ferramentas de uso diário empacotadas e prontas para invocar.
Instalação
npx convez-ai ai-native-engineer/astack1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Gerenciar e pesquisar histórico de sessões Claude Code e Codex
- Transcrever e resumir voice memos e gravações com pipeline automático
- Explorar repositórios open-source e encontrar issues para contribuição
- Coletar contexto consolidado de múltiplas fontes (Obsidian, Notion, Slack, Google Workspace)
- Automatizar testes e auditorias de performance web com Chrome DevTools
- Gerenciar campanhas de anúncios Meta (Facebook/Instagram) via CLI
Veredito

Conjunto técnico e bem pensado para fluxos nativos de IA, especialmente em ambientes macOS e multifonte. Requer dependências externas moderadas, mas oferece real ganho de produtividade para quem vive em Claude Code e precisa orquestrar workflows complexos sem sair do terminal.
README
A tradução para o português deste README ainda está na fila.
astack
Aiden's Stack — a small, curated set of Claude Code skills I actually use every day, packaged as a plugin.
The skills respond in Korean by default (they grew out of Korean voice/notes workflows), but the mechanics work in any language.
Install
/plugin marketplace add ai-native-engineer/astack
/plugin install astack@astack
Then invoke a skill with the astack: prefix, e.g. astack:session-history.
Skills
| Skill | What it does | External deps |
|---|---|---|
session-history | Unified view & search of Claude Code (~/.claude) + Codex (~/.codex) sessions — list, timeline, full-text grep, show, token usage summary. | Python 3 |
voice-memos | Apple Voice Memos / call recordings / Apple Notes / Caret MCP → transcribe, correct, search, summarize, notify. Includes a launchd watcher pipeline (auto transcribe → summarize → notify on new recordings, incl. call .m4a) with FDA diagnostics. | macOS, Python 3, apple-stt, ffmpeg, (optional) Caret MCP, Telegram/Discord |
imessage | Read & search macOS Messages (iMessage/SMS/RCS) via readonly SQLite (decodes attributedBody); send via osascript. MCP-free, on-demand. | macOS, Python 3 (stdlib only), Full Disk Access |
chrome-devtools-cli | Drive headless Chrome from the terminal via chrome-devtools-mcp's standalone CLI — navigate, click/fill, screenshot, console/network inspect, JS eval, Lighthouse audit, performance trace (Core Web Vitals), heap snapshot. On-demand alternative to the MCP server. | Node.js, chrome-devtools-mcp (npm i -g chrome-devtools-mcp@latest), Chrome/Chromium |
meta-ads-cli | Manage Meta (Facebook/Instagram) ads from the shell via the meta CLI — campaign/adset/ad/creative CRUD, performance insights (spend/CTR/CPC/ROAS, breakdowns), pixel conversion tracking, catalogs. Mental model + setup/commands/recipes references; full options delegated to meta --help. | PyPI meta-ads (uv tool install meta-ads --python 3.13), a Meta system-user access token |
crawl | Crawl web pages & doc sites to clean markdown, fully local with no API key — single page, deep-crawl to one file, or mirror a whole doc tree to per-page files (URL path = file path). Built-in traps for locale explosion, boilerplate, and broken links. | Python 3, crawl4ai crwl CLI (uv tool install crawl4ai && crawl4ai-setup), headless browser |
oss-explore | Discover-first GitHub toolkit (gh CLI): find open-source projects by topic (merged keyword+topic search, star/activity/language, with per-repo good-first-issue/help-wanted counts auto-shown), then drill into issue-level entry points (beginner-label synonyms, excludes already-PR'd/stale), browse trending ∩ contributable, bootstrap fork→clone→branch, and retrospect merged-PR contributions (org vs external OSS, star-ranked, badges) + stats (merge-rate, year/month/weekday, language). Terminal tables or dark/light HTML. Topic/language/filters are all user args — no hardcoded domain. | gh (authenticated), jq |
project-context-gather | Topic-keyword sweep across your personal/collab tools (Obsidian, voice memos / call recordings / Apple Notes / Caret, Notion, Slack, Google Workspace) → one consolidated, relevance-gated context archive per source under ./context/, with originals preserved in attachments/. Re-runs upsert into the existing per-source archive via YAML frontmatter anchor (incremental merge, dedup, date bump) instead of overwriting; scripts/context_status.py lists collection status. Subagent or main-thread modes; no web research. | rg, plus whichever per-source tools you use (agent-slack, Notion ntn CLI, gog, voice-memos skill, Caret MCP) |
html-explainer | Single-file HTML explainers with a verified visualization stack: Mermaid v11 + ELK auto-layout diagrams (text DSL, no manual coordinates → no overlaps), Apache ECharts 6 charts (built-in dark theme, log-axis range bars via custom series), Iconify icons, auto dark/light re-render. Animations are comprehension-only (edge-flow, state toggles — no decorative fade-ins). Ships a wired template, a curated pitfall guide (license traps, dead libs, AI-hallucination patterns), and a headless render-verify script run before opening. | chrome-devtools-cli skill's CLI (for verify.sh), internet for CDNs (jsDelivr) |
data-go-kr | Work-unit access to any Korean public-data API (data.go.kr): search the catalog → application guide → call, caching each successful call as a recipe (NTS business status, G2B bid/award info, …) with official reference docs preserved alongside. Calls hit the API directly with your own key — no proxy. | DATA_GO_KR_API_KEY (free data.go.kr account; per-API 활용신청), curl, Python 3 |
company-context-research | Pre-outreach / diligence company research → a per-company evidence package: fragmented public-surface map (legal entity / parent / brand / B2B / careers / IR / CDN), crawl-first recursive crawl with full first-party attachment recovery, external press inventory (Naver News API + Google News RSS yearly windows with original-URL decoding; Tavily supplement), and jurisdiction-aware official data (Korean DART called directly + listing/public-data via data-go-kr; SEC EDGAR for US-listed). | Python 3, crawl skill (crwl CLI), data-go-kr skill, Naver Developers app keys (NAVER_CLIENT_ID/NAVER_CLIENT_SECRET), (optional) DART key API_K_DART, (optional) Tavily CLI tvly |
tts | Local text-to-speech & voice cloning (Qwen3-TTS via mlx-audio, fully offline) — register a reference voice once from any audio clip, then synthesize long narration in full (one-pass) or chunk (per-sentence, individually regenerable) mode with tail-fade/pad, end-cutoff audit+regen, and a loudness-normalized edit track for video. macOS say fallback for quick non-cloned speech. | macOS (Apple Silicon), mlx-audio (uv tool install mlx-audio), huggingface_hub (hf, model weights), ffmpeg |
voice-memos setup
voice-memos is macOS-only and needs a few extras:
apple-stt(macOS SpeechAnalyzer) for transcription;ffmpegfor.qtafiles- Python deps:
cd skills/voice-memos && uv sync - Optional notifications: copy
skills/voice-memos/.env.example→skills/voice-memos/.envand fill in your Telegram/Discord values. Everything else works without it; only sending is skipped. - Optional automation: register a launchd LaunchAgent that runs
scripts/run.shon new recordings — setup, log reading, and Full Disk Access diagnostics inskills/voice-memos/references/watcher.md.
Notes
- Early cut (
v0.1.0). Somevoice-memosdocs reference~/.claude/skills/voice-memos/...paths that assume a symlinked install; running purely as a plugin may need path tweaks for the helper scripts. - More skills will be added after this MVP.
License
MIT © Seungwon An (Aiden) — see LICENSE.