Install any skill in seconds. Free to start, no credit card required.
Get Started Free →[omh] Policy overlay for browser tasks - add auth, confirmation, and observed-trace gates after preferring the native browser for ordinary URL, click, login, and form actions. Use when the user says: browser-operator, browser operator, browser task, browser operation, browser automation, browser session, webpage operation, web page operation.
.claude/skills/rlaope-omh-browser/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-22 | ✗→✓ | ▲ Improved | 181% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 91% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 105% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 206% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 11% | 0% |
This is a Hermes-native browser-operator workflow skill.
browser-operator exists so Hermes users can ask for this workflow in chat and receive a structured, evidence-bounded OMH operating surface instead of ad hoc narration.
Good example:
prepare_browser_operator_card with required context, wrapper actions, and not-evidence boundaries.Bad example:
achievements, workspace-audit, production-audit, automation-blueprint, github-event-ops, buzz, agent-board, gateway-intent-card, +34 more) - schedules, status, health, and ops review.oh-my-hermes or name the adjacent workflow.omh-routing/references/skill-common-rail.md.Use when Hermes should prepare or supervise a browser/page interaction request such as opening a URL, clicking, logging in, filling forms, or capturing page blockers without claiming browser execution.
Strong routing signals: browser-operator, browser operator, browser task, browser operation, browser automation, browser session, webpage operation, web page operation, open url, open the url, open page, open the page, visit url, visit page, navigate url, navigate page, click page, click this page, click button, click login, login page, fill form, fill the form, submit form, checkout url, capture blockers, page blockers, interactive page, browser trace, browser observation, playwright task, 웹페이지, 웹 페이지, 브라우저, 브라우저 작업, 브라우저 조작, 페이지 열고, url 열고, 링크 열고, 클릭, 로그인, 로그인 폼, 폼 작성, 폼 입력, 캡처, 막히는 부분
Category: browser Phase: browser-task Hermes role: guide Quality tier: workflow-surface-gated Reasoning demand: standard
Quality bar:
Handoff policy:
Keep this as Hermes-facing orchestration guidance first. Prepare executor, connector, gateway, or host-runtime handoff only when the user accepts that next step and observed evidence can be recorded.
Required inputs:
Expected outputs:
Artifact expectations:
Safety rules:
Preferred harness for this skill: browser-operator.
shomh runtime record --skill browser-operator --harness browser-operator --status started
Record observed delegation results; otherwise return not_available or not_observed. Prepared OMH routing is not execution, review, CI, merge-readiness, or merge evidence.
Preserve workflow intent and stop conditions; verify before claiming completion.
Use Hermes-native subagent/delegation features when available: native subagents -> Hermes delegation when available, otherwise sequential lanes.
Shared product, compatibility, topology, memory, harness, and execution rules: omh-routing/references/skill-common-rail.md. Load it when applicable; otherwise name an unavailable capability.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-22 | fail→pass | 6,111 | 6,502 | +6% | 1 | 1 | 0% | 854 | 2,400 | +181% | 0 | 0 | — |
case-01 | fail→pass | 8,678 | 8,783 | +1% | 1 | 1 | 0% | 1,533 | 2,930 | +91% | 0 | 0 | — |
case-02 | fail→pass | 20,490 | 8,192 | -60% | 1 | 1 | 0% | 1,334 | 2,735 | +105% | 0 | 0 | — |
case-03 | fail→pass | 5,828 | 7,843 | +35% | 1 | 1 | 0% | 852 | 2,603 | +206% | 0 | 0 | — |
case-04 | fail→pass | 14,191 | 9,914 | -30% | 1 | 1 | 0% | 2,762 | 3,066 | +11% | 0 | 0 | — |
case-05 | fail→pass | 7,552 | 10,852 | +44% | 1 | 1 | 0% | 1,312 | 3,054 | +133% | 0 | 0 | — |
case-06 | fail→pass | 7,641 | 7,691 | +1% | 1 | 1 | 0% | 1,297 | 2,585 | +99% | 0 | 0 | — |
case-07 | fail→pass | 12,905 | 6,541 | -49% | 1 | 1 | 0% | 2,274 | 2,509 | +10% | 0 | 0 | — |
case-08 | fail→pass | 8,748 | 8,359 | -4% | 1 | 1 | 0% | 1,429 | 2,780 | +95% | 0 | 0 | — |
case-09 | fail→fail | 2,612 | 7,757 | +197% | 1 | 1 | 0% | 330 | 2,626 | +696% | 0 | 0 | — |
case-10 | fail→pass | 13,096 | 12,466 | -5% | 1 | 1 | 0% | 521 | 2,680 | +414% | 0 | 0 | — |
case-11 | pass→pass | 11,339 | 5,900 | -48% | 1 | 1 | 0% | 1,975 | 2,365 | +20% | 0 | 0 | — |
case-12 | fail→pass | 10,718 | 5,095 | -52% | 1 | 1 | 0% | 574 | 2,154 | +275% | 0 | 0 | — |
case-13 | pass→pass | 11,344 | 6,737 | -41% | 1 | 1 | 0% | 1,812 | 2,481 | +37% | 0 | 0 | — |
case-14 | fail→pass | 9,977 | 6,160 | -38% | 1 | 1 | 0% | 1,760 | 2,325 | +32% | 0 | 0 | — |
case-15 | fail→pass | 20,275 | 8,794 | -57% | 1 | 1 | 0% | 3,566 | 2,744 | -23% | 0 | 0 | — |
case-16 | pass→pass | 17,875 | 9,187 | -49% | 1 | 1 | 0% | 3,289 | 2,871 | -13% | 0 | 0 | — |
case-17 | fail→pass | 4,562 | 5,103 | +12% | 1 | 1 | 0% | 609 | 2,282 | +275% | 0 | 0 | — |
case-18 | pass→pass | 5,961 | 6,650 | +12% | 1 | 1 | 0% | 922 | 2,425 | +163% | 0 | 0 | — |
case-19 | fail→pass | 17,916 | 9,722 | -46% | 1 | 1 | 0% | 2,997 | 3,048 | +2% | 0 | 0 | — |
case-20 | pass→pass | 5,490 | 8,701 | +58% | 1 | 1 | 0% | 879 | 2,828 | +222% | 0 | 0 | — |
case-21 | pass→fail | 10,843 | 6,111 | -44% | 1 | 1 | 0% | 1,721 | 2,405 | +40% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +64 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.