Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+21 moreClaude CodeCursorCodex CLIMuseDots
Manus
Grok
Grok Bot+18 agentsAI Image Generator: On-Brand Images, No API Keys
AI Image Generator makes your agent work like an art director with a production pipeline.
Free
AI Image Generator: On-Brand Images, No API Keys
Example session with this skill installed
Create 3 Instagram post images and a YouTube thumbnail for our new matte black travel mug. Brand colours: black and burnt orange #E8590C. Thumbnail text: "KEEPS COFFEE HOT 12H".
- Read your context and instructions
- Compiled the ai image generator
Plan
- Instagram 4:5 (1080x1350) x3 + YouTube 16:9 (1280x720)
- Drafts with FLUX Schnell (8 images, ~$0.03), finals with Gemini Flash Image (4 images, ~$0.21).
Approved by you before generating.
Style block (reused for all 4)
product photography, matte black stainless travel mug, warm burnt-orange accents (#E8590C), soft directional window light, shallow depth of field, clean minimal set, natural textures, no text
Prompts
Desk: "{style}, mug on a light oak desk next to a laptop and notebook, morning light from the left, 50 mm, negative space at top for headline" (4:5)
2. Outdoors: "{style}, mug held in a gloved hand on a misty mountain trail, steam rising, 85 mm" (4:5)
3. Flat lay: "{style}, top-down flat lay of the mug with coffee beans, a passport and a burnt-orange scarf" (4:5)
4.
Thumbnail: "{style}, mug close-up with steam, dramatic contrast, dark background, subject on the left, empty right third" (16:9)
Review
| Draft | Issue | Fix |
|---|---|---|
| Desk v1 | lid shape wrong (looked like a cup) | added "sealed flip lid, cylindrical, 450 ml" |
| Trail v2 | extra finger on the glove | changed pose to "mug resting on a rock" |
| Flat lay v1 | clutter | removed passport, more negative space |
Finals & exports
python scripts/fit_image.py desk.png --preset ig-portrait --logo logo.png -o ig-1-desk.jpg
python scripts/fit_image.py trail.png --preset ig-portrait --logo logo.png -o ig-2-trail.jpg
python scripts/fit_image.py flatlay.png --preset ig-portrait --logo logo.png -o ig-3-flatlay.jpg
python scripts/fit_image.py thumb.png --preset youtube-thumb --text "KEEPS COFFEE\nHOT 12H" --text-pos center -o yt-thumb.jpg
Text added as an overlay (sharp and correctly spelled) instead of generated inside the image.
Delivered: 4 images (180-420 KB), prompts, alt text (e.g. "Matte black travel mug on an oak desk in soft morning light"), total cost $0.24. Want a 9:16 story version of the trail shot?
Connects securely to your tools. The creator never sees your data.
About this skill
Images you can actually use, not just generate
Most AI image prompts produce one random picture with garbled text, the wrong aspect ratio and no brand consistency. AI Image Generator makes your agent work like an art director with a production pipeline.
What it does
- Short brief: purpose, placement, platform size, subject, style, brand colours and logo, text, number of variants.
- No API keys on Agensi: uses Agensi image tools - FLUX Schnell for cheap, fast drafts (under half a cent per image) and Gemini Flash Image for higher-quality finals that handle text better. Your agent shows the price and asks before each use. Also works with OpenAI, Gemini, FLUX via fal/Replicate or local Stable Diffusion.
- Expert prompting: subject, composition, lens, lighting, style, palette and mood instead of "beautiful, 8k"; aspect ratio chosen at generation time; recipes for product shots, lifestyle, ads, thumbnails, blog headers, hero images, icon sets and textures.
- Consistent sets: a reusable style block (palette, lighting, medium) so every image in a campaign or icon set matches.
- Draft -> select -> final: 2-4 drafts reviewed against the brief (details, hands, text, artifacts, brand colours, room for text), targeted re-prompts, then the final render.
- Finishing script (
fit_image.py, tested): smart crop or pad to 11 presets (Instagram, story, LinkedIn, X, OG image, YouTube thumbnail, Pinterest, hero, shop, favicon), bold text overlays, logo placement, optimised JPG/WebP/PNG. - Delivered with: the exact prompts, model and cost, alt text, and variation ideas.
- Safe & honest: no real people, no copied characters, logos or living artists' styles, no misleading product images, AI disclosure where required.
Includes
- SKILL.md workflow
- scripts/fit_image.py - export/crop/overlay tool (Pillow)
- references/prompting.md - prompt template, 8 recipes, style block, fixes for common problems, alt-text formula
Works with Claude Code, Claude (Cowork), Codex, Cursor and any SKILL.md-compatible agent.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
Text overlays are best kept under 5 words. Cannot generate identifiable people, copyrighted characters, or violent content. Final output depends on the specific AI model's capabilities.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
This skill has been used 8 times.
Trust & safety
Security scan
Security scan passed 2 days ago
- Free with an Agensi account