Clarify proof surfaces and simplify operator usability

[Reedtrullz]

Summary

• Add a proof-surface model and CLI that distinguishes current-main proof, committed dashboard snapshots, and local generated readbacks.
• Split local-app smoke/golden-path helpers mechanically into agent_os.local_app_smoke.
• Collapse below-the-fold /today, /resume, and Goal evidence into closed secondary details sections while preserving proof fields, hashes, counters, non-claims, and first-viewport affordances.
• Add report-only artifact hygiene classification.
• Add a static local read-only hosted-dashboard export with explicit zero-effect counters.

Verification

Local:

• python3 -m pytest tests/test_proof_surface.py tests/test_local_app_smoke.py tests/test_operator_surfaces.py tests/test_artifact_hygiene.py tests/test_hosted_dashboard_export.py -q -> 14 passed
• python3 -m compileall -q agent_os tests
• python3 -m agent_os.cli --root "$scratch" app-golden-path-smoke-test -> app_golden_path_smoke_test: passed
• python3 -m agent_os.cli proof-surface
• python3 -m agent_os.cli artifact-hygiene
• python3 -m agent_os.cli hosted-dashboard-export
• git diff --check

GitHub Actions on head 0321021b1edcc07535a5615c0621795e7749ae03:

• Run 28881424357 succeeded.
• Fast Smoke job 85670371854 succeeded.
• Full Pytest Suite job 85671060981 succeeded.

Proof State

Current branch proof surface reports:

• live_proof_state: branch_same_sha
• merge_claim: branch_proof_only
• latest_ci_run_id: 28881424357
• latest_ci_scope: workflow_run
• latest_ci_status: success

Non-claims

• This PR does not claim current-main proof until the merge commit lands on main and a successful full workflow run is recorded for that exact SHA.
• No deploy, provider call, remote worker, browser/desktop adapter, scheduler, or external mutation is included in this slice.
• Generated .clanker readbacks remain local proof/readback artifacts unless intentionally promoted.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.