Slide Images
SkillDocs & knowledgeAdds a slide Claude skill so your agent can find, generate, and insert images into presentation slides.
Use Slide Images in Claude, ChatGPT or Ahel Desktop
Free. Sign in, add Slide Images and connect your AI. About a minute.
Also: Claude Code · Cursor · Codex
Then ask your AI: use the Slide Images skill
Details
Instructions available. Your AI can read the instructions. Execution depends on the setup they require.
Account requirements not reviewed. Check the skill instructions before use; ahel provides instructions and does not run this skill.
No other account needed.
Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.
About this skill
Source and generate images for slides with Assets-owned style grounding and provenance. Use when a slide needs a new visual, logo, or approved media.
What this skill tells your AI
The instructions your AI receives, as published by builderio/agent-native in templates/slides/.agents/skills/slide-images/SKILL.md and read by ahel’s review.
Images for slides are generated or sourced through the Slides actions and the Assets app. The runtime agent must use the Assets-grounded path described below; the local CLI is only a developer entry point.
When a chat attachment has an embeddable URL, add it to the current slide with
update-slide using <img src="…">, then call view-screen and re-read the
slide with get-deck to confirm it persisted.
Scripts
| Script | Purpose | Example |
|---|---|---|
generate-image | Local helper for the Assets-grounded generation action | pnpm action generate-image --prompt "hero image" --count 3 |
search-images | Search Google Images via the configured provider | pnpm action search-images --q "Acme logo transparent" --count 5 |
search-logos | Resolve company domains and canonical logo URLs | pnpm action search-logos --q "Acme" |
image-gen-status | Check configured image providers | pnpm action image-gen-status |
Image Generation Flow
For agent or editor generation, call generate-image-api, not a provider or
legacy image action directly:
- Resolve the deck's active design system, its
imageStyle, the current slide role, and any approved Creative Context references. - Call
generate-image-apiwith the prompt plus bounded deck and slide context. Assets chooses the library, preset, style anchors, model, and fallback behavior while preserving provenance. - Show each returned variation as an inline rendered preview using markdown
image syntax (
), not a plain link ([Variation 1](url)) — the chat renders![]()as an actual image but[]()as a bare link. - Preserve the returned
assetId,runId,previewUrl, anddownloadUrl. - For direct insertion, call
generate-image-apiwithinsertIntoSlide: true,deckId, andslideId. Only say the image was added when it returnsinserted: true; a preview URL is not proof of a slide write. - For preview-only variations, place the chosen URL with
update-slide, then re-read the deck withget-deckusingcompact=falseand confirm the persisted HTML contains the image source before reporting success. - For feedback, refine the same asset rather than starting an unrelated run.
Context to pass
The image brief should include the visual role, subject, composition/crop,
format, must-preserve content, and exclusions. Pass deckId, slideId, and
slideContent so Assets can ground the result in the actual slide. Do not use
generic style references to override a linked design system or preset.
search-images and search-logos remain bounded lookup tools for finding
existing media. They are not a substitute for the active design system or a
style-generation brief.
Logo Lookup
Two options for company logos:
Option 1: canonical logo search (uses Logo.dev search when configured and a bounded domain fallback otherwise):
pnpm action search-logos --q "Acme"
Use a returned logoUrl directly. Do not call a second logo-provider action for
each result.
Option 2: Google Image Search (fallback):
pnpm action search-images --q "Acme logo transparent" --count 5
Important Rules
- Use the active design system, library, preset, and approved style anchors for visual consistency; do not invent a second style language in the prompt
- Use
.fmd-img-placeholderdivs in slides before real images are generated - Use one canonical provider action per conceptual search; do not loop legacy provider scripts or manually guess provider URLs
- Never claim image insertion from a generation response alone; require the
verified
inserted: trueresponse, or anupdate-slideplusget-deckcompact=falseverification for preview-only placement
Signals
- GitHub stars
- 7k
- Forks
- 635
- Last commit
- Oct 2026
Advanced
- Item type
- skill
- Key
slide-images- Source
- github.com/builderio/agent-native
github.com/builderio/agent-native
More in Docs & knowledge
Skill · mattpocock
More in Docs & knowledgecanvas-design
Skill · anthropics
More in Docs & knowledgedoc-coauthoring
Skill · anthropics
More in Docs & knowledgewriting-for-agents
Skill · mattpocock
More in Docs & knowledgespec-driven-development
Skill · addyosmani
More in Docs & knowledgedefuddle
Skill · kepano
More in Docs & knowledge