Security audit
Security checks across malware telemetry and agentic risk
The skill’s image-generation purpose is real, but it gives the agent broad, partly hidden setup and credential-handling authority that users should review before installing.
Install only if you are comfortable sending prompts and any chosen reference images to WeryAI, storing an IMAGE_GEN_API_KEY locally, and letting the agent manage setup files. Prefer a project-scoped API key with limited billing exposure, avoid sensitive local images unless you intend to upload them, review any tool-install prompt carefully, and rotate/remove the key if the workspace is shared or committed.
When the user wants image generation but no model is configured (`EXTEND.md`, `--model`, and `IMAGE_GEN_DEFAULT_MODEL` are all absent): **Do not ask the user to read docs or edit config files.** Follow the guided flow in [references/config/first-time-setup.md](references/config/first-time-setup.md) § "Model Selection — Agent-Guided Flow": 1. Start from the bundled starter registry shipped with this skill. If the model list looks stale or a requested model is missing, refresh it silently with `npm run discover-image-models -- --out .image-skills/image-generation/MODELS.json` 2. Initialize local defaults with **Nano Banana 2** (`GEMINI_3_1_FLASH_IMAGE`) by writing `EXTEND.md`
These rules apply to every user interaction. They are not optional guidelines. 1. **Never show commands, file paths, config syntax, or schema details to the user.** All tool execution is silent. 2. **Never ask the user to edit config files.** The agent writes `EXTEND.md`, `MODELS.json`, and other configs on behalf of the user. 3. **Ask one question at a time.** Do not present all dimensions at once. Lead with the highest-impact choice. 4. **Offer concrete options with descriptions**, not raw technical names. If presenting styles, describe what each looks like. 5. **Always disclose the model and estimated time** when generation starts.
2. **Propose installation in user-friendly terms.** Say something like: "I need to set up a small runtime to generate images — OK if I install it now? Takes about 30 seconds." 3. **On user approval**, run `npm install -g bun` or the platform-appropriate installer silently. Report success or failure in plain language. 4. **If the user declines or installation fails**, explain honestly: "I can't generate images in this environment right now. You can ask your admin to set up the image toolchain, or try again in a different environment." 5. **Never ask the user to install tools themselves** or paste terminal commands. ## Troubleshooting
Treat the API key as a secret: - prefer writing it locally on the user's behalf instead of making the user edit files - do not ask the user to paste the key into normal chat if a local write path is available - if the user does share the key, never echo it back or include it in follow-up messages ## 2. API Key
3. **If `IMAGE_GEN_API_KEY` is missing**, ask the user for permission to configure it now 4. Only after readiness and API key are resolved, continue to model selection / prompt clarification / generation **Access token**: only use **`IMAGE_GEN_API_KEY`**. It may also live in `.image-skills/image-generation/.env` as `IMAGE_GEN_API_KEY=...`. After the user approves, the agent may persist it there by running `npm run setup -- --project . --workflow <workflow> --persist-api-key` when the key is already in env, or by writing the file locally on the user's behalf instead of asking the user to edit files manually. **First-use readiness check**: before the first generation run in a new OpenClaw or local instance, the agent must run:
No VirusTotal findings
Detected: suspicious.dangerous_exec, suspicious.env_credential_access, suspicious.potential_exfiltration
