菜单
Find & remove duplicate photos using perceptual hashing. Interactive browser review page — no server needed. Use when user says: dedup photos, find duplicate photos, remove duplicate images, photo dedup, /photo-dedup.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Build premium, TV-commercial-grade product highlight-reel promo videos from a handful of screen recordings. Music-only by default (optional VO-narrated cut), HyperFrames + paused GSAP, rendered and muxed locally. Ships a full kit — 16:9 + 9:16 films and matching key-art posters — in three swappable, production-proven themes. Use when the user asks for a product promo / launch film / hero reel / app commercial / "video like <reference>" / feature sizzle from screen captures, or says "highlight reel", "product TV commercial", "aurora theme", "minimal / notion-style promo". Three theme presets: AURORA (dark, glowing, chromatic, bold), MINIMAL (light Notion-white, floating cards, one accent, confident restraint), and LINE ART (cream paper, bold navy ink, one gold accent, hand-drawn framed panels). Ships a self-contained EnConvo scaffold (bundled logo, keycaps, SFX, music bed) that reproduces the launch film by dropping in new product clips.
Turn screen recordings into polished videos — marketing promos, user guides, product demos, and more. Goal-aware pipeline: detects user intent, selects strategy, recommends a plan, then executes. Full pipeline: intent detection → strategy selection → source analysis → storyboard planning → source prep → VO generation → frame-by-frame compositing → audio mixing → final encode. Supports animated presenters (AI animal/character with rembg cutout), per-word caption sync (pop, karaoke, static styles), multi-zoom animations, overlay dissolve transitions, time-mapped VO-to-source sync, CJK-aware captions, letterbox-aware cropping, and keypress + shortcut animation overlays with frame-aligned tick/tock SFX (canonical pattern for product demos that show hotkeys; ships proven SFX library at assets/sfx/). Use when: (1) user has screen recordings and wants a polished video — marketing, tutorial, demo, or changelog, (2) user says "make a promo video", "tutorial from this recording", "TikTok video", "marketing video", "use
Unified financial analysis & portfolio management toolkit. Routes to the right sub-system based on intent. Covers: daily market briefings, macro risk monitoring, sentiment scanning, strategy evolution & signals, options analysis, 禅動 chart analysis, COT smart-money reports, and full portfolio checking (structured notes, sell puts, FX options, accumulators) with XLSX report generation. Use when user says /financial-analysis, 'daily briefing', 'morning briefing', 'macro risk', 'liquidity check', 'sentiment scan', 'market sentiment', 'evolve strategy', 'create strategy', 'strategy signal', 'what does the strategy say', 'quick look', 'scan ticker', 'ride the rally', 'catch dips', 'play safe', '禅動', 'chandong', 'analyze ticker', 'fcn check', 'fcn status', 'check fcn', 'portfolio check', 'portfolio report', 'structured products', 'are my fcns safe', 'sell puts', 'accumulators', 'options analysis', 'cot report', 'smart money'.
Visible step-by-step walkthrough mentor for live browser/app tutoring across OptionCharts.io, TradingView, and EnConvo Settings. Use when the user wants to be walked through a page, chart, setup flow, tool, or concept viewport by viewport, including options/stock analysis with option chains, GEX, DEX, open interest, max pain, Greeks, volatility, P&L charts, TradingView charts, indicators, drawing tools, watchlists, screeners, alerts, layouts, paper trading, or EnConvo setup tasks such as global providers, AI models, credentials, agents, tools, skills, shortcuts, dictation, knowledgebase, account, and developer settings.
Conduct rigorous, McKinsey-style industry / sector research and produce a structured research report. Use when the user says /industry-research-mckinsey or asks to research an industry/sector/market/value chain, wants market sizing (TAM/SAM/SOM), competitive landscape, value-chain / industry-chain analysis, profit pools, demand-supply dynamics, industry drivers, or trend/outlook analysis, or asks for a 行业研究 / 产业研究 / 行业报告 / 赛道分析 / 投研报告 / industry report / sector deep-dive / market analysis. Also fires when the user wants to find supply bottlenecks, pricing-power chokepoints, or the source of excess returns (超额收益) along a value chain (the 产业投资人 / bottleneck lens). Bilingual: follows the user's language. Do NOT use for company-only financial modeling unrelated to an industry, or for pure document formatting.
Produce multi-act long-form Mandarin financial analysis videos for social-media upload (TikTok/Bilibili/YouTube/WeChat视频号/X). Given any input source (news article, earnings call, IPO filing, macro print, chart, tweet thread, raw data), build a broadcast-grade anchor-led explainer using a locked pipeline — script in tight 5字/second cadence, render anchor with xAI Grok Imagine Video i2v with baked Mandarin VO and locked identity, composite editorial GSAP overlay panels via HyperFrames in the right 40% of frame, concat acts, mix BGM, ship MP4. Use when the user gives a finance/markets/macro/equity/options/IPO/crypto source and asks for a viral video, a 解读视频, a 拆解视频, an explainer, a TG/SNS-ready clip, or simply says "make a video about this". Also use when the user says "finance-capital-markets-video-maker", "FCM video", "market video", "anchor video", or "Vivieen video".
| name | photo-dedup |
| description | Find & remove duplicate photos using perceptual hashing. Interactive browser review page — no server needed. Use when user says: dedup photos, find duplicate photos, remove duplicate images, photo dedup, /photo-dedup. |
| user_invocable | true |
Use this skill when the user wants to deduplicate photos, find unique images, remove similar/duplicate photos. Trigger phrases: "dedup photos", "find duplicate photos", "remove duplicate images", "photo dedup", "/photo-dedup".
Processes a folder of photos, identifies duplicates and near-duplicates using perceptual hashing, and generates an interactive HTML review page. The user reviews duplicate groups and picks which to keep — no server required.
Perceptual Hashing — Each image is converted to a perceptual hash (pHash). Similar-looking images produce similar hashes, even if they differ in resolution, compression, or minor edits.
Clustering — Images are grouped by hash similarity. Each cluster = one "scene". The best image (largest file = highest quality) is auto-selected.
Interactive Review — A self-contained HTML page opens in the browser. The user reviews duplicate groups, picks which to keep, and downloads a save script.
/photo-dedup ~/Photos/school-event/
/photo-dedup ~/Photos/school-event/ --threshold 8
Threshold controls similarity sensitivity (default: 6, range 0-20). Lower = stricter, higher = more aggressive grouping.
When the user invokes this skill:
pip3 install Pillow imagehash pillow-heifpython3 ~/.claude/skills/photo-dedup/scripts/dedup.py <source_folder> --preview [--threshold N]
http://localhost:54535/enconvo/delete_files) to delete unselected duplicate files directlyNo server needed for the review page. The HTML file is self-contained — works offline, opens instantly. File deletion is handled via the Enconvo local API.
<source>/photo_dedup.html ← Review page (open in browser)
/tmp/dedup_report_*.json ← Scan report