Is the Draw Image skill free?

Yes. The skill itself is free from ai-sns/ai-sns. SkillProof publishes the install command and an independent test verdict at no cost.

How do I install Draw Image?

Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.

Can I use Draw Image with Cursor, Copilot, Gemini CLI, Codex or other AI tools?

The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.

Draw Image

Text-to-image via OpenAI images API, then uploads the result to gofile.io public hosting

by ai-sns · ai-sns/ai-sns

Tested · Didn't pass

Read why it failed Download source ↗

Test report

Verdict: Tested · Didn't pass
Tested: Jul 31, 2026
Environment: Claude Code 2.x (agent harness)
Upstream re-checked: Aug 10, 2026 · cdb10b5

Installed the openclaw copy into a throwaway HOME and ran the documented command verbatim; it returned {"ok": false, "error": "API HTTP 401: Incorrect API key provided: your_api_key"}, disproving SKILL.md's claim that a key is built in and that no configuration is needed. I had no OpenAI image key, so I could not generate an image and produced no skill artifact to compare against a baseline (outputMeasured=false). Reading scripts/draw.py I found ssl.CERT_NONE on all three network calls and confirmed it by execution: pointing --api-base at https://self-signed.badssl.com with --api-key sk-secret-test completed the TLS handshake and returned an HTTP 404 body, while a plain urllib.request.urlopen against the same host raised CERTIFICATE_VERIFY_FAILED — so the Bearer key is sent over an unverified channel. Trigger phrasings judged: "Draw me a picture of a cow in a spacesuit" (load), "Generate an image of a futuristic city at sunset" (load), "Can you paint a watercolor landscape for my blog header?" (load), "Create a flowchart diagram of our auth flow" (do not load — that is mermaid/SVG, not text-to-image), "Resize and compress these PNGs in ./assets" (do not load — manipulation, not generation); 5/5 correct.

Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works

Installs cleanly 2/5
Triggers reliably 5/5
Output vs. baseline 0/10
Docs & honesty 2/5

What Draw Image does

Wraps a Python script that calls an OpenAI-compatible /v1/images/generations endpoint (default model gpt-image-1-mini) with a prompt, size, quality and count, saves the PNG locally, then uploads it to the anonymous file host gofile.io and returns only the public download-page URL. Intended to trigger when a user asks to draw, paint, generate or create an image. It is a sample skill from the AI-SNS agent-network project rather than a Claude Code skill, and requires your own OpenAI API key despite documentation claiming a key is built in.

How to install Draw Image

git clone --depth 1 https://github.com/ai-sns/ai-sns.git /tmp/draw-image-src
mkdir -p ~/.claude/skills
cp -R /tmp/draw-image-src/adapters/openclaw/skill_example/draw-image ~/.claude/skills/draw-image
# Requires python3 (stdlib only) AND your own OpenAI API key with image-generation access.
# SKILL.md claims "The script has a built-in API base and key, so no extra configuration is
# needed" -- this is FALSE. The shipped DEFAULT_API_KEY is the literal string "your_api_key"
# and OpenAI returns HTTP 401. You must set: export DRAW_IMAGE_API_KEY=sk-...
# (optionally export DRAW_IMAGE_API_BASE=https://your-openai-compatible-host/v1)
# SECURITY, read before using: scripts/draw.py sets ssl.CERT_NONE + check_hostname=False on
# every request, including the one carrying "Authorization: Bearer <your key>". Verified by
# execution: it completed a handshake with https://self-signed.badssl.com while stock
# urllib refused the same host with CERTIFICATE_VERIFY_FAILED. Your API key is MITM-able.
# Also: every generated image is unconditionally uploaded to gofile.io (anonymous public file
# sharing) with no opt-out flag, and only that public URL is returned.
# Run: python3 ~/.claude/skills/draw-image/scripts/draw.py --prompt "a cow" --size 1024x1024
# (SKILL.md's own examples hardcode C:\Users\IDD\.openclaw\workspace\skills\... and will not run.)
# Two other copies exist in the repo: adapters/hermes/skill_example/ai-sns/draw-image (identical
# script, ~/.hermes paths) and aisns_backend/skills/draw-image (NOT Claude Code compatible --
# non-standard frontmatter with runner/skill_key, and its draw.py reads creds from a local
# SQLite DB belonging to the AI-SNS backend).

Skills live in ~/.claude/skills/ (global) or .claude/skills/ (per-project). Restart Claude Code after installing.

Commands — how to trigger Draw Image

/draw-image Text-to-image via OpenAI images API, then uploads the result to gofile.io public hosting

It also activates on plain-language prompts like these:

Draw a sunset over a mountain lake
Generate an illustration of a robot barista
Create an image of a futuristic city skyline

Frequently asked questions

Is the Draw Image skill free?: Yes. The skill itself is free from ai-sns/ai-sns. SkillProof publishes the install command and an independent test verdict at no cost.
Does Draw Image work with Claude Code?: We tested it with Claude Code 2.x (agent harness) on Jul 31, 2026. Verdict: Tested · Didn't pass. Installed the openclaw copy into a throwaway HOME and ran the documented command verbatim; it returned {"ok": false, "error": "API HTTP 401: Incorrect API key provided: your_api_key"}, disproving SKILL.md's claim that a key is built in and that no configuration is needed. I had no OpenAI image key, so I could not generate an image and produced no skill artifact to compare against a baseline (outputMeasured=false). Reading scripts/draw.py I found ssl.CERT_NONE on all three network calls and confirmed it by execution: pointing --api-base at https://self-signed.badssl.com with --api-key sk-secret-test completed the TLS handshake and returned an HTTP 404 body, while a plain urllib.request.urlopen against the same host raised CERTIFICATE_VERIFY_FAILED — so the Bearer key is sent over an unverified channel. Trigger phrasings judged: "Draw me a picture of a cow in a spacesuit" (load), "Generate an image of a futuristic city at sunset" (load), "Can you paint a watercolor landscape for my blog header?" (load), "Create a flowchart diagram of our auth flow" (do not load — that is mermaid/SVG, not text-to-image), "Resize and compress these PNGs in ./assets" (do not load — manipulation, not generation); 5/5 correct.
How do I install Draw Image?: Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
Can I use Draw Image with Cursor, Copilot, Gemini CLI, Codex or other AI tools?: The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.