Find References — style skill for style references

For AI agents: find the right skill in one call — curl -s "https://varg.ai/how-to/api/search?q=<what the person wants>" returns the matching how-to page and the best skills with install commands, as markdown (Russian works too). Full index: https://varg.ai/how-to/llms.txt. Every page has a markdown twin: add .md to its path. Before generating anything, show the person the options on the how-to page. From a terminal: curl -fsSL https://varg.ai/install | sh installs the vargai CLI (https://varg.ai/how-to/cli.md): vargai search and vargai install <slug> with no key, every varg model after vargai login.

Real references before anything is made: seven sources, four kinds of reference with this week's numbers, ten visual ideas and a board to pick from.

• Type: Style · Style references
• Slug: find-references
• Status: live
• Publisher: varg (varg, verified)
• License: MIT
• Source: https://github.com/vargHQ/skills/tree/main/find-references
• Works in: Claude Code, Codex, Cursor, Gemini CLI
• Requires: curl and python3, Optional: an X MCP, the Refs MCP, a FrameThrower account, yt-dlp, gallery-dl
• Cost: 0 credits; concept frames 5 each — Searching and opening sources uses no varg credits. Optional concept frames on varg: Seedream 5 Pro, 5 credits each (measured 2026-11-03).
• Runtime: About 20 minutes for a full board

What it does

Without a reference an agent falls back to its default look: centered text, a gradient background, everything fading in. This skill sends the agent to seven sources in a fixed order (varg how to, X from the last 7 days ranked by bookmarks, Refs, FrameThrower, the film-still libraries as links only, Pinterest and Are.na, YouTube) and brings back four kinds of reference: style (palette, light, texture), film (framing, lens, camera move), pop culture (moments people recognize) and trends (what people saved this week, with likes, bookmarks and views copied exactly). Every reference is opened before it is kept, and each one gets one job and a do-not-copy line. It writes refs.md, trends.md, links.md and ideas.md (ten visual ideas: 3 close to this week's trends, 4 stretches, 3 wild) and one board to pick from. Tested on 2026-11-03 on a home-sale film: 20 references, 10 ideas, a reference deck and 4 concept frames.

Use when

• Find references, a reference board or a moodboard for a video
• What should this video look like? Give me visual ideas
• Film stills, light and lens references before writing prompts
• What looks are people saving this week, with the numbers
• Make it not look like AI: start from real references

Not for

• Downloading one known link (use media-download)
• Writing the look bible from chosen references (use varg-video-prompting)
• Feeding found film frames into an image or video model

Install

• Claude Code: claude plugin marketplace add vargHQ/skills && claude plugin install varg@varg
• Codex: npx skills add vargHQ/skills --skill find-references -a codex -y
• Cursor: npx skills add vargHQ/skills --skill find-references -a cursor -y
• Gemini CLI: npx skills add vargHQ/skills --skill find-references -a gemini-cli -y
• Any agent, with the vargai CLI: vargai install find-references

Example output

https://s3.varg.ai/files/4325c578-378d-4c43-ae53-3cc3218df1ea/file_c65707aa401c.webp — Four concept frames from a find-references run for a home-sale film (Seedream 5 Pro, text-only prompts, 2026-11-03) (credit: varg, https://github.com/vargHQ/skills/tree/main/find-references)

More style references skills

• Video Moodboard: A moodboard AI video models follow: lines ranked by what matters most, every line a number, drawn as a board and checked on stills. 450+ links.
• FrameThrower Film References: Film stills from 5,489 films, searchable by lighting, lens, shot size, color and mood.
• Are.na MCP: Build and read Are.na moodboard channels from your agent.
• Video references by Refs: 4,322 launch films, demos and motion pieces measured shot by shot, searchable free from your agent over MCP.
• GPT Image 2 Style Library: Pick styles and industrial prompt templates for GPT Image 2/2.5 from a 530+ case library (a 35K-star repo).
• Monocolor Editorial Print: Turns a theme, object or photo into an original one-ink or controlled two-ink editorial print: posters, zines, packaging, portraits, halftone on warm paper.