fix(browser): pass sandbox flags via AGENT_BROWSER_ARGS

[LeonSGP43]

What does this PR do?

Fixes the sandbox-bypass env handoff in tools/browser_tool.py so local browser launches pass Chromium flags through the env var that agent-browser actually reads.

Today Hermes detects when --no-sandbox --disable-dev-shm-usage is needed, but writes those flags to AGENT_BROWSER_CHROME_FLAGS. agent-browser expects AGENT_BROWSER_ARGS, so the bypass never reaches Chrome and headless launches still fail on root/AppArmor-restricted hosts.

Related Issue

Fixes #23496

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 🔒 Security fix
• 📝 Documentation update
• ✅ Tests (adding or improving test coverage)
• ♻️ Refactor (no behavior change)
• 🎯 New skill (bundled or hub)

Changes Made

• switch the sandbox bypass env injection from AGENT_BROWSER_CHROME_FLAGS to AGENT_BROWSER_ARGS in tools/browser_tool.py
• add regression coverage in tests/tools/test_browser_homebrew_paths.py for both root and AppArmor-restricted launches
• assert we no longer emit the stale AGENT_BROWSER_CHROME_FLAGS env var in the launch path

How to Test

1. Run uv run --frozen pytest -q -o addopts='' tests/tools/test_browser_homebrew_paths.py
2. Confirm the new tests cover root sandbox bypass and AppArmor userns restriction detection
3. Optionally reproduce on a Linux/AppArmor host and verify browser commands now launch with AGENT_BROWSER_ARGS=--no-sandbox --disable-dev-shm-usage

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: macOS 26.4.1

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — or N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide — or N/A
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

Screenshots / Logs

• uv run --frozen pytest -q -o addopts='' tests/tools/test_browser_homebrew_paths.py -> 24 passed
• uv run --frozen ruff check tools/browser_tool.py tests/tools/test_browser_homebrew_paths.py -> passed

[alt-glitch]

Competing PRs for #23496: #23532, #21659, #20014, #22582. All fix the same AGENT_BROWSER_CHROME_FLAGS → AGENT_BROWSER_ARGS rename.

[shan-dot-ee]

See note from original issue #23496 :

• A simpler in-place rename of AGENT_BROWSER_CHROME_FLAGS to AGENT_BROWSER_ARGS within the current block was not sufficient to restore launches in our testing; we'd encourage maintainers to reproduce on an Ubuntu 24.04 AppArmor host before considering a minimal rename as the fix.

[teknium1]

Automated hermes-sweeper review: this browser sandbox env-var fix is already implemented on main.

Evidence:

• tools/browser_tool.py:2042 now writes the auto-injected sandbox flags to AGENT_BROWSER_ARGS as --no-sandbox,--disable-dev-shm-usage.
• Commit 8ed2ef6f46e9642acfba57b4b8da893a574ecfd0 made the core AGENT_BROWSER_CHROME_FLAGS → AGENT_BROWSER_ARGS fix and called out the AppArmor / "No usable sandbox" failure mode.
• Commit 4695d2716f60da89152bdc9dfa7d96e54ea7c22e followed up by honoring user-provided AGENT_BROWSER_ARGS and documenting the behavior.
• The linked issue [Bug]: Headless VM fails to use browser: tools/browser_tool.py writes AGENT_BROWSER_CHROME_FLAGS but agent-browser reads AGENT_BROWSER_ARGS --sandbox-bypass injection is a no-op #23496 includes a later reporter comment saying it was fixed in 8ed2ef6.
• The fix is contained in release tag v2026.5.16.