tiktok-carousel

Generates a 6-slide TikTok carousel with images and text, creates a draft post via Postiz API, and outputs a caption for review and publishing.

This skill's code appears to do what it claims (generate images/captions and optionally upload drafts to Postiz), but the registry metadata failing to declare required env vars (OPENAI_API_KEY and POSTIZ_API_KEY) is an inconsistency you should address before installing. Before use: 1) Verify the skill's provenance — ask the publisher for a homepage or repo and confirm you trust it. 2) Confirm the Postiz service (api.postiz.com) is legitimate and review its API docs; if uncertain, do not provide your production POSTIZ_API_KEY. 3) Use limited-scope or temporary API keys and avoid reusing high-privilege keys. 4) Run first in an isolated environment (VM/container) and monitor outbound network requests (or run under a local HTTP proxy) to observe what endpoints are contacted. 5) If you won't use Postiz, remove/disable postiz_api_integration.py or avoid running scripts/upload.py. 6) Request that the registry entry be updated to list OpenAI and Postiz credentials explicitly and provide a homepage/source for auditing — that would raise confidence from suspicious to benign.

Type: OpenClaw Skill Name: tikto-automation Version: 1.0.0 The skill bundle is designed to generate images and upload them to an external service (Postiz), which requires file system and network access. While its core functionality aligns with the description, the `postiz_api_integration.py` and `scripts/upload.py` modules accept arbitrary file paths for upload, and `tiktok_content_gen.py` accepts arbitrary paths for local file writes. If an AI agent were to call these scripts with unsanitized user input, it could lead to arbitrary file upload or local file write vulnerabilities, allowing an attacker to potentially exfiltrate arbitrary files or overwrite system files. There is no clear evidence of intentional malicious behavior (e.g., hardcoded exfiltration of sensitive files, backdoors, or explicit prompt injection against the agent to perform harmful actions), but the broad file access capabilities without explicit input sanitization make it suspicious.