Bug: BrowserAutomationSkill is a stub — tools return queued status, never execute

[mrveiss]

Problem

autobot-backend/skills/builtin/browser_automation.py defines a BrowserAutomationSkill with three tools (scrape_page, fill_form, take_screenshot) but the implementation is a stub — tools return "queued" status without actually executing browser actions.

Discovered During

Design exploration for #1967 (Web Pipeline Engine).

Impact

Severity: Medium

The skills framework advertises browser automation capability that doesn't work. Agents attempting to use the skill via the skills system get no-op results. The same functionality works through browser_mcp.py tools directly, so this is a dead code path rather than a missing capability.

Expected Behavior

Either:

• a) Wire the skill tools to call playwright_service methods (matching what browser_mcp.py already does), or
• b) Remove the stub and rely on browser_mcp.py as the sole browser tool interface (simpler)

Notes

With #1967 adding pipeline capabilities to browser_mcp.py, option (b) may be cleaner — avoid two parallel browser tool interfaces.

[mrveiss]

Fixed in PR #2484 (merged). Removed BrowserAutomationSkill stub — browser_mcp.py is canonical.