
from clawra2,284
Edit a fixed reference image with Grok Imagine and send AI-edited selfies to messaging channels (Discord, Telegram, WhatsApp, Slack) via OpenClaw.
This skill edits a fixed reference image using xAI's Grok Imagine (Fal.ai) to produce AI-generated 'selfies' and then distributes the resulting images to messaging platforms using the OpenClaw CLI. It provides prompt modes (mirror vs direct) for outfit/full-body or close-up portrait edits, automates prompt construction, calls the image edit API, extracts the resulting image URL, and sends it to the requested channel(s).
Use this skill when a user requests an image of the agent (e.g., 'send me a pic', 'send a selfie wearing a santa hat') or when you need to programmatically generate and distribute stylized avatar/selfie images across chat platforms. It's ideal for fun social interactions, marketing posts, or automated art distribution.
This skill is implementable by agents with Bash/HTTP and OpenClaw CLI access; works well with agents that can call external HTTP APIs (Fal.ai/Grok Imagine) and run OpenClaw commands to post media.
Clawra Selfie is a niche skill for generating AI-edited selfies of a fixed character image using xAI's Grok Imagine model via fal.ai, then sending them to messaging channels via OpenClaw. The bash script has solid error handling and colored logging but exited immediately due to missing FAL_KEY (expected in DRY_RUN). The TypeScript script was skipped (unknown extension). Minor security concerns: $CHANNEL/$CAPTION interpolated into JSON without jq escaping in the bash fallback path, and the TS script passes untrusted strings into execAsync creating a command injection vector.
@fal-ai/client (optional, falls back to fetch)ts-node (for TS script)Fun but very niche skill. Fixed reference image on jsDelivr CDN means it's specific to one character. The bash script's JSON construction in the fallback curl path doesn't use jq for $CHANNEL and $CAPTION, unlike $PROMPT which is properly escaped. The TypeScript execAsync command construction is vulnerable to shell injection. No malicious intent detected — just standard API integration code with minor injection risks.