Selenium Browser

Start a Selenium‑controlled Chrome browser, open a URL, take a screenshot, and report progress. Supports headless mode and optional proxy.

Do not install or run this skill as-is. Key concerns: (1) The SKILL.md, the script embedded in SKILL.md, and the actual scripts on disk disagree: one version takes and reports a screenshot, another never takes a screenshot and loops forever. (2) SKILL.md says the skill returns JSON, but the embedded script prints a Python dict (not serialized JSON) and the on-disk script prints nothing — the agent will likely fail to parse results. (3) SKILL.md promises a hardcoded save path (/home/main/clawd/diffusion_pdfs/) that isn't implemented; confirm where files will be written and make that path configurable. Actions you should take before using this skill: inspect and reconcile the actual scripts (which file will be executed), require the executable to output valid JSON, add a timeout or controlled termination instead of an infinite loop, ensure screenshots are written to a safe, configurable directory, run the skill in an isolated container or sandbox with limited permissions, and test with non-sensitive URLs. Also prefer a publisher-provided homepage or source repository (currently missing) and ask the author to fix the inconsistencies and provide a clear termination/protocol for the process. If you cannot validate and fix these issues, treat the skill as untrusted.

Type: OpenClaw Skill Name: selenium-browser Version: 1.0.0 The skill is classified as suspicious due to a significant functional discrepancy between its documentation (SKILL.md) and its actual implementation (scripts/launch_browser.py). While SKILL.md describes taking a screenshot and reporting its path, the Python script instead enters an infinite loop (`while True: time.sleep(1)`) after navigation, keeping the browser alive indefinitely without taking a screenshot or producing the expected JSON output. This leads to unexpected behavior and potential uncontrolled resource consumption. Additionally, the `webdriver.Chrome` constructor in `scripts/launch_browser.py` incorrectly passes `executable_path=chrome_bin` (the browser binary) when a `Service` object (already configured with `chromedriver_path`) is also provided, indicating a bug that could prevent the skill from functioning correctly.