Generate/edit images via Tuzi API (default), Google Gemini, OpenAI, DashScope, Replicate. Text-to-image + image-to-image editing; 1K/2K/4K resolution. Use for image create/modify/edit requests incl. --input-image.

Generate/edit images via Tuzi API (default), Google Gemini, OpenAI, DashScope, Replicate. Text-to-image + image-to-image editing; 1K/2K/4K resolution. Use fo...

This skill largely matches its stated purpose, but proceed cautiously. Specific things to consider before installing/running: - Metadata mismatch: the registry lists no required env vars while SKILL.md/code require many provider API keys; verify you trust this code before supplying secrets. - The skill reads ~/.tuzi-skills/.env and ./ .tuzi-skills/.env and merges values into process.env. Check those files for unrelated secrets before running or avoid placing secrets there. - The code uses child_process.execSync with interpolated values to call 'identify' (ImageMagick) and to run curl when an HTTP proxy is set. If you provide proxy strings, API keys, or file paths with unexpected characters this could cause command-line issues or injection. Prefer running in a controlled environment and inspect the code yourself. - SKILL.md preflight only checks for npx; you may also need ImageMagick ('identify') and curl available depending on options — the script falls back if 'identify' fails, but behavior differs. - If you want to reduce risk: run the script locally in an isolated environment (container or VM), review the providers' base URLs in code, and avoid storing unrelated secrets in the .tuzi-skills/.env files. If you plan to use this skill in production/automated agents, ask the author to fix the metadata (declare required env vars) and to avoid shelling out with unescaped values (use HTTP libraries instead of exec+curl and child_process-safe calls).

Type: OpenClaw Skill Name: tuzi-nano-banana Version: 1.0.0 The skill bundle contains multiple critical shell injection vulnerabilities. In `scripts/main.ts`, the `autoDetectResolution` function executes the `identify` command using `execSync` with an unsanitized file path from the `--input-image` argument. Similarly, `scripts/providers/google.ts` uses `execSync` to run `curl` with unsanitized proxy settings derived from environment variables. The script also automatically loads environment variables from `.tuzi-skills/.env` files in the current working directory or home directory, which increases the risk of exploitation. While these appear to be unintentional coding flaws rather than deliberate malware, they present a high risk of Remote Code Execution (RCE) if the agent is provided with malicious input or run in a compromised directory.