moonlight-lupin/agent-skills

pexels-stock-photos

Search and download free real-world stock photos from the Pexels API.

ソースを見る
リポジトリの原文

見出し、例、コード、表、リンク、参照画像を含む原文を表示しています。

Pexels Stock Photos

Search and download free stock photos from the Pexels API (200 req/hr, 20k/month free). All content is free for commercial and non-commercial use; just credit the photographer and link to Pexels.

Routing: Pexels vs AI Generation

User says...Tool
"find a photo of Singapore" / "search for office pictures" / "stock image of a desk"Pexels (this skill)
"I need a real photo for this slide" / "find pictures for my article"Pexels (this skill)
"generate an image of a cat" / "create a picture of..." / "make me an illustration"Host's image generation tool
"upscale this" / "remove background" / "edit: make warmer"Host's AI image editing skill
"production-ready hero image" / "let's do this properly"Host's multi-stage image generation skill

The key word test: if the user says photo, picture, stock image, or find/search → Pexels. If the user says generate, create, make, draw, or illustration → AI generation.

When to Use

  • User asks for a "photo of X" or "stock image of Y"
  • Building presentations (PPTX) that need real imagery
  • Article/blog illustrations needing authentic photos
  • Social media posts requiring professional photography
  • Any request for "real" images vs AI-generated

Don't use for:

  • AI art / creative generation → host's image generation tool
  • Editing existing images → host's image editing skill
  • Screenshots or diagrams → use appropriate tools

API Reference

Base URL: https://api.pexels.com/v1/ Auth: Authorization: <PEXELS_API_KEY> header

Endpoints

EndpointMethodPurpose
/searchGETSearch photos by query
/curatedGETCurated trending photos
/photos/:idGETGet a specific photo by ID

Search Parameters

ParamTypeDefaultNotes
querystringrequirede.g. "singapore skyline"
orientationstringlandscape, portrait, square
sizestringlarge (24MP), medium (12MP), small (4MP)
colorstringred, orange, blue, ... or hex #ffffff
localestringen-USzh-CN, ja-JP, etc.
per_pageint15Max 80
pageint1Pagination

Photo Resource src Sizes

SizeDimensionsUse case
originalFull resPrint, high-quality
large2x940×650 @2xRetina displays
large940×650Web, presentations
medium~350px heightThumbnails, previews
small~130px heightSmall thumbnails
portrait800×1200Vertical/mobile
landscape1200×627Slides, social
tiny280×200Placeholders

Workflow

1. Search for photos

bash
curl -s -H "Authorization: $PEXELS_API_KEY" \
  "https://api.pexels.com/v1/search?query=singapore+skyline&per_page=5&orientation=landscape"

Parse the JSON response to present results to the user: photo ID, photographer, alt text, dimensions, and Pexels URL.

2. Download a photo

Use the `src` URLs from the search response — don't hand-construct images.pexels.com URLs (the slug/extension isn't guaranteed; Pexels' docs say to use src). Each photo object carries ready-made size variants (see the src table above).

bash
# Download the size you need, straight from the photo's src field
# (e.g. the value of photos[0].src.large from the search JSON)
curl -sL -o /tmp/pexels_<id>.jpeg "<photo.src.large>"

Deliver the file to the user via the host's file delivery mechanism.

3. Present results

When showing search results, format as a table:

| # | ID | Photographer | Alt | Dimensions |
|---|-----|-------------|-----|------|
| 1 | 5097071 | Dylan Chan | Singapore skyline at dusk | 6016×3384 |

Include the Pexels URL so the user can preview before downloading.

Attribution

Per Pexels guidelines:

  • Credit the photographer: "Photo by [Name] on Pexels"
  • Link back to the photo page URL
  • Include "Photos provided by Pexels" when displaying multiple

Example attribution block:

Photo by Dylan Chan on Pexels
https://www.pexels.com/photo/high-rise-buildings-in-singapore-5097071/

Common Pitfalls

  1. API key not in environment. PEXELS_API_KEY must be set before making

requests. Load it from your environment or .env file before calling curl.

  1. Using spaces in query. URL-encode spaces as + or %20.

singapore skylinesingapore+skyline.

  1. Downloading original for previews. Original files can be 10MB+. Use

large or medium for previews, original only when the user needs full resolution.

  1. Rate limit awareness. 200 req/hr, 20k/month — but only API calls

(search, curated, photo lookup) count. CDN downloads from the src URLs on images.pexels.com (no API key needed) do not count. Batch searches with per_page=15-80 to minimize API calls.

  1. Not including attribution. Pexels requires visible credit to the

photographer and a link to Pexels. Always include attribution text when delivering photos.

  1. Hand-built photo URLs. Don't construct images.pexels.com/photos/...

URLs yourself — the slug and extension aren't guaranteed across photos. Always download via the src.* URLs from the API response.

Verification Checklist

  • [ ] PEXELS_API_KEY is set in the environment
  • [ ] Search query returns results (check total_results > 0)
  • [ ] Downloaded file exists and is non-empty
  • [ ] Attribution text included with delivered photo
  • [ ] Rate limit headers checked if making many requests (X-Ratelimit-Remaining)
同じリポジトリから

関連する Skills

すべての Skills
moonlight-lupin
コミュニティ

claude-plugin-converter

Convert Claude Code plugins into self-contained Hermes plugins — discovery analysis then full conversion

導入数
1
GitHub Stars
62
更新日
9月7日
moonlight-lupin
コミュニティ

clips-studio

Use when the user wants to "make a video/clip", "generate a video from text", "animate this image/photo", "create a marketing reel / social video / teaser", "do a 3D / parallax move", "pan/zoom/orbit a shot", or mentions fal.ai, Kling, Veo or Seedance for video. Staged fal.ai workflow — brainstorm, draft cheaply, produce the final at quality. Three modes: text-to-video, animate (still to motion), camera-move (push-in/pan/orbit). Not for still images (image-studio), slide decks, or charts. Honesty discipline: never depict real identifiable subjects via text-to-video.

導入数
1
GitHub Stars
62
更新日
9月7日
moonlight-lupin
コミュニティ

decision-log

ADR-style decision journal for agents and teams. Create numbered decision records, track superseding chains, schedule periodic reviews, and search past decisions to avoid re-litigating settled questions.

導入数
1
GitHub Stars
62
更新日
9月7日
moonlight-lupin
コミュニティ

deep-research

Autonomous multi-step deep research engine implementing an iterative Think → Search → Extract → Synthesize → Stop loop. The LLM drives every decision: what to search, what's relevant, what's missing, and when to stop. Produces a cited, magazine-quality report with inline citations, category- specific formatting, and research stats. Trigger when the user asks for "deep research", "research report on", "comprehensive analysis of", "look into X in depth", "write a report on X", or any question needing multi-source synthesis beyond a single search. For entity vetting/dossiers use entity-research; for news digests use news-monitoring; for source-grounded Q&A use notebooklm-mode.

導入数
1
GitHub Stars
62
更新日
9月7日