POST AI Automation

Automate TikTok and Instagram video creation and scheduled posting from one product image using POST AI with customizable styles, captions, and performance t...

This skill appears to implement the advertised video-generation and upload workflow, but several issues need attention before you run it: - Expect to provide sensitive credentials: POST AI API key and TikTok/Instagram cookie files/session_id. The code reads them from config.json (plaintext). Prefer storing secrets in environment variables or a secure keyring rather than a repo file. - Registry metadata did NOT declare these credentials or Python requirements — that's an inconsistency. Treat the package as requiring Python 3.8+ and manual credential configuration. - SKILL.md references scripts that are missing (previews.py, daily_post.py, track_performance.py). The package has placeholders (TODO) for actual TikTok/Instagram upload logic. The upload functions currently simulate behavior; real browser automation or API calls would need additional dependencies and credentials. - config.example.json contains a JSON syntax error (malformed posting_schedule entry). Fix the example before copying to config.json. - Review how cookies/session_id are obtained and stored. Exporting full browser cookies can expose other accounts; only extract the minimum tokens needed and store them securely. - Audit the code paths that download images (arbitrary URLs) and run subprocesses — run in an isolated environment (container or VM) and avoid running with elevated privileges. - Verify the legitimacy of POST AI (postai.myscalev.com / api.postai.com) and costs before entering payment/keys. If you decide to use it: run it in a sandbox, fix the config example, move secrets out of repo files, and manually inspect/complete the upload implementation (or replace with a vetted library) before scheduling automatic runs.

Type: OpenClaw Skill Name: postai-automation Version: 1.0.0 The skill is classified as suspicious primarily due to a Server-Side Request Forgery (SSRF) vulnerability in `scripts/batch_process.py`. The `download_image` function uses `urllib.request.urlopen(url)` where the `url` is directly taken from the `image_url` column in the input CSV, allowing an attacker to potentially make the agent request arbitrary internal network resources or local files. Additionally, the skill handles highly sensitive credentials like social media session IDs and cookie file paths (e.g., `tiktok_cookies.json`) which, while necessary for its stated purpose, presents a significant attack surface if exploited. There are no clear indicators of intentional malicious behavior like data exfiltration to unauthorized endpoints or prompt injection against the agent.