@omniaibot/win-x64 1.1.2 → 1.3.0

package/bin/omnibot-windows-x64/VERSION

@@ -1 +1 @@
-1.1.2
+v1.3.0

package/bin/omnibot-windows-x64/omnibot/skills/omnibot/SKILL.md

@@ -5,21 +5,21 @@ compatibility: Requires omnibot v2 CLI daemon and the omnibot Chromium extension
 allowed-tools: Bash
 metadata:
   author: omnibot
-  version: "2.1.0"
+  version: "2.2.0"
 ---
 
 # Omnibot
 
 Omnibot is Browser Infrastructure for AI Agents.
 
-It connects Hermes, Claude Code, Codex, OpenCode, MCP Agent Runtime, and other agent systems to a real Chromium browser through the local omnibot daemon and CLI.
+It connects Hermes, Claude Code, Codex, OpenCode, and other agent systems to a real Chromium browser through the local omnibot daemon and CLI.
 
 This skill is an execution specification for agents. It is not a human command manual.
 
 ## When to Use
 
 - Use Omnibot when browser runtime state matters: login, cookies, storage, client-side rendering, extensions, or visible user tabs.
-- Use Omnibot when an agent must read rendered pages, click controls, fill forms, upload files, navigate, extract content, or collect evidence.
+- Use Omnibot when an agent must read rendered pages, click controls, fill forms, navigate, extract content, or collect evidence.
 - Do not use Omnibot when static files or plain HTTP content already answer the task.
 
 ## Core Rules
@@ -35,16 +35,16 @@ This skill is an execution specification for agents. It is not a human command m
 - Parse JSON output from commands that return JSON.
 - Do not claim success without verification evidence.
 
-## Session + Tab Double Lock
+## Workflow Context + Tab Target
 
 Every page-state workflow must use both controls:
 
 - `OMNIBOT_SESSION_TOKEN=<workflow-name>` for workflow isolation.
 - `--tab-id <TAB_ID>` for page targeting.
 
-They are not substitutes. Session controls which workflow owns state. Tab ID controls which page receives the command.
+OMNIBOT_SESSION_TOKEN is a workflow/context token, not a browser session and not a tab target. It isolates agent state such as refs, trace, recording, aliases, and temporary resources. Tab ID controls which page receives the command.
 
-Do not rely on default tab, active tab, current tab, or prior targeting state. Every page command requires explicit `--tab-id`.
+Do not rely on default tab, active tab, current tab, or prior targeting state. Every page-state command requires explicit `--tab-id` except when the command is only discovering or creating a tab.
 
 Use the same token inside one workflow. Use different tokens for independent workflows.
 
@@ -56,7 +56,7 @@ OMNIBOT_SESSION_TOKEN=research omnibot click --tab-id <TAB_ID> @e4
 OMNIBOT_SESSION_TOKEN=research omnibot snapshot -i --tab-id <TAB_ID>
 ```
 
-`@eN` refs are tab-scoped. never reuse `@eN` refs across tabs.
+`@eN` refs are tab-scoped. Refs rule: never reuse `@eN` refs across tabs.
 
 The agent dispatch pattern is token + tab-id:
 
@@ -75,6 +75,14 @@ OMNIBOT_SESSION_TOKEN=checkout omnibot dom dblclick n1 --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=checkout omnibot clipboard read --tab-id <TAB_ID>
 ```
 
+Targeting categories:
+
+| Command category | Rule |
+| --- | --- |
+| Page-state commands: `snapshot`, `click`, `fill`, `type`, `get`, `is`, `wait`, `screenshot`, `console`, `network`, `execute-js`, `dom`, `mouse`, `clipboard`, `viewport`, `assets`, `goto`, same-tab `navigate` | Require `--tab-id <TAB_ID>`. |
+| Discovery or tab creation: `tabs`, `tab list`, `tab new <URL>`, `open <URL>`, `navigate <URL>` | Return or discover the tab id; save it for later page-state commands. |
+| Read URL exception: `read <URL>` | Opens a temporary read tab and does not need `--tab-id`; reading an existing tab still requires `read --tab-id <TAB_ID>`. |
+
 ## Observe -> Act -> Verify
 
 All page operations must follow this loop:
@@ -104,22 +112,22 @@ If verification fails, do not repeat blindly. Re-observe, choose the next fallba
 
 ## Pattern > Command
 
-Start from the task, not the command name. Use operation patterns for read, click, fill, select/check, navigation, wait, extraction, upload, and batch.
+Start from the task, not the command name. Use operation patterns for read, click, fill, select/check, navigation, wait, extraction, and batch.
 
 Use fallback tiers only after standard patterns fail. Collect evidence separately from fallback execution. The command reference is only a lookup table; it must not decide behavior.
 
-## Read vs Scan Routing
+## Read vs Snapshot Routing
 
 Choose by intent before choosing by command name:
 
 | Intent | Prefer | Why |
 | --- | --- | --- |
 | Summarize or extract page content, article text, search results, feeds, or long/lazy pages | `read` | Returns clean text/Markdown for agent reasoning. |
-| Quickly observe the current viewport before deciding what to do next | `scan --json` | Returns structured viewport state useful for routing. |
+| Quickly observe current visible UI structure before deciding what to do next | `snapshot -i` | Returns actionable refs and page structure for routing. |
 | Find buttons, links, inputs, or refs for a later action | `find` or `snapshot -i` | Produces actionable targets; `read` does not. |
 | Verify one concrete condition | `get`, `is`, or `wait` | Narrow evidence is more reliable than dumping a page. |
 
-Do not use `scan` as a long-form reader. Do not use `read` as the first step for click/fill workflows unless the user explicitly asked for page content first.
+Do not use `read` as the first step for click/fill workflows unless the user explicitly asked for page content first.
 
 ## Quick Routing
 
@@ -133,7 +141,6 @@ Do not use `scan` as a long-form reader. Do not use `read` as the first step for
 | Navigation | `references/operation-patterns.md#navigation` |
 | Waiting | `references/operation-patterns.md#wait` |
 | Extraction | `references/operation-patterns.md#extraction` |
-| Upload | `references/operation-patterns.md#upload` |
 | Batch | `references/operation-patterns.md#batch` |
 | Fallback | `references/fallback-operations.md` |
 | Tabs and sessions | `references/session-and-tabs.md` |
@@ -182,7 +189,7 @@ Use screenshots, annotated screenshots, console logs, network logs, trace, recor
 - Missing `OMNIBOT_SESSION_TOKEN` or missing `--tab-id` on page-state commands.
 - Acting without verify or claiming success without evidence.
 - Using `execute-js` first, raw CSS before semantic find, `@eN` across tabs, implicit tab targeting, or shell sleep instead of `omnibot wait`.
-- Using removed commands: `switch-tab`, `focus-tab`, `tab switch`. These are no longer available.
+- Using removed commands: `switch-tab`, `focus-tab`, `tab switch`, `tab focus`. These are no longer available.
 - Using `tabs[0]` or the first tab as a target. The first tab is often a user tab or transport tab.
 - Closing user tabs discovered during cleanup. Only close tabs created by the current workflow.
 

package/bin/omnibot-windows-x64/omnibot/skills/omnibot/references/anti-patterns.md

@@ -4,7 +4,7 @@
 | --- | --- | --- |
 | Missing `OMNIBOT_SESSION_TOKEN` | Workflows can share implicit state across agents. | `OMNIBOT_SESSION_TOKEN=research omnibot snapshot -i --tab-id <TAB_ID>` |
 | Missing `--tab-id` | The command may target the wrong page. | Pass `--tab-id <TAB_ID>` on every page-state command. |
-| `execute-js` first | JavaScript bypasses standard browser interaction and hides better evidence. | Try semantic find, snapshot refs, selectors, DOM, and mouse first. |
+| `execute-js` first | JavaScript bypasses standard browser interaction and hides better evidence. | Try semantic find, snapshot refs, selectors, DOM, and mouse first. For CSS-only dropdowns, use JavaScript only after the trigger clicked, options remain absent from `snapshot -i`, and higher tiers cannot complete open-then-select. |
 | Reusing `@eN` across tabs | Refs are tab-scoped and may point to a different element elsewhere. | Take a fresh `snapshot -i --tab-id <TAB_ID>` per tab. |
 | Relying on implicit tab targeting | It is unavailable and unsafe for concurrent workflows. | Use explicit `--tab-id` on every page-state command. |
 | Using sleep instead of wait | Fixed sleeps are flaky and either too short or too slow. | Use `omnibot wait --text`, `--url`, selector state, load state, or `--fn`. |
@@ -12,7 +12,6 @@
 | Claiming success without evidence | The browser may not have changed as expected. | Cite verified URL, text, element state, screenshot, console, or network evidence. |
 | Raw CSS before semantic find | CSS is more brittle and less tied to user intent. | Start with `find role/text/label/placeholder/testid`. |
 | Screenshot when text extraction is enough | Screenshots cost more and are harder to parse. | Use `read` for page content or `get text` for a selector. |
-| Using `scan` for long-form reading | `scan` is viewport observation and may miss long/lazy content. | Use `read --screens N --tab-id <TAB_ID>`. |
 | Using `read` before an action workflow | `read` returns text, not stable action targets. | Use `find`, `snapshot -i`, `get`, or `is` for click/fill/verify workflows. |
 | Headless expecting existing user login | Headless does not automatically share the user's visible browser session. | Use visible/background mode for existing login or explicitly log in headless. |
 | Mixing multiple workflows in one session token | Agents can overwrite each other's default state. | Use different tokens: `research`, `checkout`, `debug`. |
@@ -31,5 +30,8 @@
 - "The ref probably still points to the same element."
 - "A screenshot is enough even though I need text."
 - "Sleep should be fine."
+- "The trigger clicked, so another `find text` should eventually select the option."
+- "`snapshot -i` cannot see the option, so the element must not exist."
+- "This dropdown is custom, so I can start with JavaScript."
 
 All of these mean: stop, re-enter the standard pattern, and make state explicit.

package/bin/omnibot-windows-x64/omnibot/skills/omnibot/references/command-reference.md

@@ -2,6 +2,8 @@
 
 This is a lookup table, not a behavior guide. Choose behavior from `operation-patterns.md`, use `session-and-tabs.md` for targeting, and use `fallback-operations.md` before dropping to lower tiers.
 
+Unless marked as discovery, runtime, or tab creation, page-state examples assume `OMNIBOT_SESSION_TOKEN=<workflow>` and `--tab-id <TAB_ID>`.
+
 ## Runtime & Status
 
 ### doctor
@@ -80,30 +82,19 @@ Fallback relation: Not a fallback command.
 
 ## Reading
 
-### scan
-
-Purpose: Observe the current rendered viewport, especially before choosing a follow-up action.
-
-```bash
-OMNIBOT_SESSION_TOKEN=research omnibot scan --json --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=research omnibot scan --text-only --tab-id <TAB_ID>
-```
-
-Preferred pattern: See `operation-patterns.md#read` and `operation-patterns.md#extraction`.
-Fallback relation: Use before `dom visible` or `execute-js` when structured viewport state is needed. Use `read` instead for long-form page text.
-
 ### read
 
 Purpose: Read clean rendered page text/Markdown from an existing tab or a temporary URL tab.
 
 ```bash
 OMNIBOT_SESSION_TOKEN=research omnibot read --screens 5 --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=research omnibot read --screens 5 --timeout 120 --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=research omnibot read --screens 3 https://example.com/article
 OMNIBOT_SESSION_TOKEN=research omnibot read --json --screens 5 --tab-id <TAB_ID>
 ```
 
 Preferred pattern: See `operation-patterns.md#read` and `operation-patterns.md#extraction`.
-Fallback relation: Prefer for page content. Use `scan --json` when the next step is action planning, and `snapshot -i` when actionable refs are needed.
+Fallback relation: Prefer for page content. Use `snapshot -i` when the next step is action planning, and `get`/`is`/`wait` for narrow state verification.
 
 ### snapshot -i
 
@@ -241,7 +232,7 @@ OMNIBOT_SESSION_TOKEN=form omnibot select @e5 "US" --tab-id <TAB_ID>
 ```
 
 Preferred pattern: See `operation-patterns.md#select--check`.
-Fallback relation: Prefer refs/selectors before JavaScript value changes.
+Fallback relation: Prefer refs/selectors before JavaScript value changes. For CSS-only dropdowns whose options are absent from `snapshot -i`, see `fallback-operations.md#css-only-dropdown-javascript-fallback`.
 
 ### check / uncheck
 
@@ -281,14 +272,28 @@ Fallback relation: Mouse drag is Tier 5 if element drag fails.
 
 ### upload
 
-Purpose: Upload a local file through a file input.
+Purpose: Reserved command for local file upload. Native upload is currently unavailable in the extension transport.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=form omnibot upload "input[type=file]" /path/to/file.png --tab-id <TAB_ID>
+# Do not use in agent workflows until file chooser support is implemented.
 ```
 
 Preferred pattern: See `operation-patterns.md#upload`.
-Fallback relation: Use JavaScript only if native upload cannot reach the input.
+Fallback relation: Do not treat JavaScript as a normal upload fallback.
+
+### keyboard / keydown / keyup
+
+Purpose: Send keyboard text or key state to the target tab when element-specific `type` or `press` is insufficient.
+
+```bash
+OMNIBOT_SESSION_TOKEN=form omnibot keyboard type "hello" --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=form omnibot keyboard inserttext "hello" --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=form omnibot keydown Shift --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=form omnibot keyup Shift --tab-id <TAB_ID>
+```
+
+Preferred pattern: Prefer `fill`, `type`, and `press` first.
+Fallback relation: Keyboard commands are a focused fallback before mouse or JavaScript.
 
 ## Semantic Find
 
@@ -350,7 +355,7 @@ OMNIBOT_SESSION_TOKEN=repair omnibot execute-js --file /tmp/repair.js --tab-id <
 ```
 
 Preferred pattern: See `fallback-operations.md#javascript-fallback`.
-Fallback relation: After semantic, refs, selector, DOM, and mouse are insufficient.
+Fallback relation: After semantic, refs, selector, DOM, and mouse are insufficient. Valid examples include CSS-only dropdowns where the trigger exists but option refs are absent from `snapshot -i`.
 
 ### cdp
 
@@ -358,8 +363,8 @@ Purpose: Send raw Chrome DevTools Protocol commands.
 Usage tier: Tier 7 fallback and inspection only.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot cdp Runtime.evaluate '{"expression":"document.title"}' --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot cdp DOM.getDocument '{"depth":1}' --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot cdp Runtime.evaluate '{"expression":"document.title"}' --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot cdp DOM.getDocument '{"depth":1}' --tab-id <TAB_ID>
 ```
 
 Preferred pattern: See `fallback-operations.md#raw-cdp-fallback` and `debugging-and-evidence.md#cdp-inspection`.
@@ -369,13 +374,13 @@ Fallback relation: Last resort.
 
 ### navigate / open / goto
 
-Purpose: Navigate or open pages. These commands create or use CLI target state; record the resulting tab id and use `--tab-id` afterward.
+Purpose: Navigate or open pages. New-tab forms return a tab id. Same-tab navigation requires an explicit target tab id. There is no CLI current-tab state.
 
 ```bash
 OMNIBOT_SESSION_TOKEN=research omnibot navigate https://example.com
-OMNIBOT_SESSION_TOKEN=research omnibot navigate --same-tab https://example.com
+OMNIBOT_SESSION_TOKEN=research omnibot navigate https://example.com --same-tab --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=research omnibot open https://example.com
-OMNIBOT_SESSION_TOKEN=research omnibot goto https://example.com
+OMNIBOT_SESSION_TOKEN=research omnibot goto https://example.com --tab-id <TAB_ID>
 ```
 
 Preferred pattern: See `operation-patterns.md#navigation` and `session-and-tabs.md#tab-explicit`.
@@ -425,7 +430,6 @@ OMNIBOT_SESSION_TOKEN=research omnibot frame main
 Purpose: Wait for time, selector, text, URL, load state, or JavaScript condition.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=checkout omnibot wait 500 --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=checkout omnibot wait "#ready" --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=checkout omnibot wait "#spinner" --state hidden --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=checkout omnibot wait --text "Welcome" --tab-id <TAB_ID>
@@ -433,17 +437,18 @@ OMNIBOT_SESSION_TOKEN=checkout omnibot wait --url "/dashboard" --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=checkout omnibot wait --load domcontentloaded --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=checkout omnibot wait --load networkidle --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=checkout omnibot wait --fn "window.appReady === true" --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=checkout omnibot wait 500 --tab-id <TAB_ID>
 ```
 
 Preferred pattern: See `operation-patterns.md#wait`.
-Fallback relation: Prefer over shell sleep.
+Fallback relation: Prefer condition waits over shell sleep. Fixed-time waits like `wait 500` are diagnostic or last resort only.
 
 ### batch
 
-Purpose: Send short command arrays as JSON or from a file.
+Purpose: Send short extension/CDP command arrays as JSON or from a file. Batch is not a wrapper for normal CLI commands such as `snapshot` or `click`.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=research omnibot batch '[{"cmd":"scan"},{"cmd":"snapshot","interactive":true}]' --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=research omnibot batch '[{"cmd":"cdp","method":"Runtime.evaluate","params":{"expression":"document.title","returnByValue":true}}]' --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=research omnibot batch --file /tmp/omnibot-batch.json --tab-id <TAB_ID>
 ```
 
@@ -457,8 +462,8 @@ Fallback relation: Not a substitute for verification.
 Purpose: Capture visual evidence from a browser tab.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot screenshot --tab-id <TAB_ID> -o /tmp/omni-shot.png
-OMNIBOT_SESSION_TOKEN=debug omnibot screenshot --annotate --tab-id <TAB_ID> -o /tmp/omni-annotated.png
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot screenshot --tab-id <TAB_ID> -o /tmp/omni-shot.png
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot screenshot --annotate --tab-id <TAB_ID> -o /tmp/omni-annotated.png
 ```
 
 Preferred pattern: See `debugging-and-evidence.md#screenshot`.
@@ -469,21 +474,21 @@ Fallback relation: Evidence collection, not operation fallback.
 Purpose: Read or clear browser console evidence.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot console logs --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot console errors --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot console clear --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot console logs --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot console errors --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot console clear --tab-id <TAB_ID>
 ```
 
 Preferred pattern: See `debugging-and-evidence.md#console`.
-Fallback relation: Evidence collection, not operation fallback.
+Fallback relation: Evidence collection, not operation fallback. `console errors` currently uses the console log collection path; verify level filtering in output before treating it as error-only evidence.
 
 ### network
 
 Purpose: Read browser network evidence.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot network logs --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot network summary --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot network logs --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot network summary --tab-id <TAB_ID>
 ```
 
 Preferred pattern: See `debugging-and-evidence.md#network`.
@@ -531,18 +536,17 @@ Fallback relation: Use after page content indicates assets are needed.
 
 ### visibility
 
-Purpose: Inspect, set, or launch automation browser visibility mode.
+Purpose: Inspect or set automation browser visibility mode. `visibility launch` currently reports planned configuration only; it does not launch a browser.
 
 ```bash
 omnibot visibility status
 omnibot visibility set background
 omnibot visibility set visible
-omnibot visibility set headless --user-data-dir /tmp/omnibot-headless
 omnibot visibility launch headless --user-data-dir /tmp/omnibot-headless
 ```
 
 Preferred pattern: See `session-and-tabs.md#visibility`.
-Fallback relation: Headless does not inherit existing user login state.
+Fallback relation: Headless does not inherit existing user login state. Treat `visibility launch` status `planned` as not launched.
 
 ### browser
 
@@ -560,7 +564,7 @@ Fallback relation: Use in multi-browser or multi-runtime scenarios.
 
 ### session
 
-Purpose: Name or list sessions.
+Purpose: Name or list Omnibot workflow context state for the current `OMNIBOT_SESSION_TOKEN`. This does not switch tabs, preserve login, or replace `--tab-id`.
 
 ```bash
 OMNIBOT_SESSION_TOKEN=checkout omnibot session name "checkout"
@@ -577,11 +581,11 @@ Fallback relation: Not a fallback command.
 Purpose: Capture, replay, or trace workflows for evidence.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot record start
-OMNIBOT_SESSION_TOKEN=debug omnibot record stop -o flow.json
-OMNIBOT_SESSION_TOKEN=debug omnibot replay flow.json
-OMNIBOT_SESSION_TOKEN=debug omnibot trace start
-OMNIBOT_SESSION_TOKEN=debug omnibot trace stop -o trace.zip
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot record start
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot record stop -o flow.json
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot replay flow.json
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot trace start
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot trace stop -o trace.zip
 ```
 
 Preferred pattern: See `debugging-and-evidence.md#record-and-replay` and `debugging-and-evidence.md#trace`.

package/bin/omnibot-windows-x64/omnibot/skills/omnibot/references/css-only-dropdown-js.md

@@ -0,0 +1,50 @@
+# CSS-Only Dropdown JavaScript Recipe
+
+Use only after semantic find, refs, selectors, DOM, and mouse fallback cannot expose or select the dropdown option.
+
+Example for a visible trigger like `按留言时间排序` and a target option text:
+
+```js
+const triggerText = '按留言时间排序';
+const optionText = '目标选项文本';
+
+const visibleElements = [...document.querySelectorAll('button, [role="button"], [aria-haspopup], div, span, li')]
+  .filter((el) => {
+    const style = getComputedStyle(el);
+    const box = el.getBoundingClientRect();
+    return style.display !== 'none' && style.visibility !== 'hidden' && box.width > 0 && box.height > 0;
+  });
+
+const byExactText = (text) => visibleElements.find((el) => el.textContent?.trim() === text);
+const trigger = byExactText(triggerText);
+if (!trigger) {
+  return { ok: false, reason: 'trigger-not-found', triggerText, optionText };
+}
+
+trigger.click();
+await new Promise((resolve) => requestAnimationFrame(() => requestAnimationFrame(resolve)));
+
+const option = [...document.querySelectorAll('div, span, li, button, [role="option"], [role="menuitem"]')]
+  .filter((el) => {
+    const style = getComputedStyle(el);
+    const box = el.getBoundingClientRect();
+    return style.display !== 'none' && style.visibility !== 'hidden' && box.width > 0 && box.height > 0;
+  })
+  .find((el) => el.textContent?.trim() === optionText);
+
+if (!option) {
+  return { ok: false, reason: 'option-not-found-after-trigger-click', triggerText, optionText };
+}
+
+option.click();
+await new Promise((resolve) => requestAnimationFrame(() => requestAnimationFrame(resolve)));
+
+return {
+  ok: true,
+  triggerText,
+  optionText,
+  selectedText: trigger.textContent?.trim(),
+};
+```
+
+Verify with page state after running the script: selected label, sorted result order, result count, URL/query state, or another user-visible condition.

package/bin/omnibot-windows-x64/omnibot/skills/omnibot/references/debugging-and-evidence.md

@@ -4,13 +4,15 @@ Debugging is for evidence. Fallback is for completing operations.
 
 Use this file when you need to prove browser state, explain a failure, or collect artifacts for review. Do not use `execute-js` as the first debug tool.
 
+Use a unique workflow/debug token per task, such as `debug-checkout` or `debug-research`. Do not share one global `debug` token across independent agents.
+
 ## Screenshot
 
 Use screenshots when visual layout, canvas, dialogs, or visual regressions matter.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot screenshot --tab-id <TAB_ID> -o /tmp/omni-shot.png
-OMNIBOT_SESSION_TOKEN=debug omnibot screenshot --annotate --tab-id <TAB_ID> -o /tmp/omni-annotated.png
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot screenshot --tab-id <TAB_ID> -o /tmp/omni-shot.png
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot screenshot --annotate --tab-id <TAB_ID> -o /tmp/omni-annotated.png
 ```
 
 Prefer text extraction when text is enough. Screenshot is evidence, not extraction.
@@ -20,26 +22,26 @@ Prefer text extraction when text is enough. Screenshot is evidence, not extracti
 Use console logs for runtime JavaScript errors, frontend warnings, or app-level diagnostics.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot console errors --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot console logs --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot console clear --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot console errors --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot console logs --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot console clear --tab-id <TAB_ID>
 ```
 
-Collect console errors before and after the action when debugging regressions.
+Collect console diagnostics before and after the action when debugging regressions. `console errors` currently uses the console log collection path; verify level filtering in output before treating it as error-only evidence.
 
 ## Network
 
 Use network logs for failed requests, redirects, blocked assets, API errors, or loading stalls.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot network summary --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot network logs --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot network summary --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot network logs --tab-id <TAB_ID>
 ```
 
 Pair network evidence with page verification:
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot get text "#main" --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot get text "#main" --tab-id <TAB_ID>
 ```
 
 ## Trace
@@ -47,10 +49,10 @@ OMNIBOT_SESSION_TOKEN=debug omnibot get text "#main" --tab-id <TAB_ID>
 Use trace for a reproducible artifact around a failing workflow.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot trace start
-OMNIBOT_SESSION_TOKEN=debug omnibot click --tab-id <TAB_ID> @e4
-OMNIBOT_SESSION_TOKEN=debug omnibot snapshot -i --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot trace stop -o trace.zip
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot trace start
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot click --tab-id <TAB_ID> @e4
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot snapshot -i --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot trace stop -o trace.zip
 ```
 
 Trace is evidence. It does not replace operation verification.
@@ -60,17 +62,17 @@ Trace is evidence. It does not replace operation verification.
 Use record/replay when a flow must be captured or reproduced.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot record start
-OMNIBOT_SESSION_TOKEN=debug omnibot click --tab-id <TAB_ID> @e4
-OMNIBOT_SESSION_TOKEN=debug omnibot snapshot -i --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot record stop -o flow.json
-OMNIBOT_SESSION_TOKEN=debug omnibot replay flow.json
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot record start
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot click --tab-id <TAB_ID> @e4
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot snapshot -i --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot record stop -o flow.json
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot replay flow.json
 ```
 
 After replay, verify final state:
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot get url --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot get url --tab-id <TAB_ID>
 ```
 
 ## CDP Inspection
@@ -78,8 +80,8 @@ OMNIBOT_SESSION_TOKEN=debug omnibot get url --tab-id <TAB_ID>
 Use CDP for low-level inspection when normal evidence is insufficient.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot cdp Runtime.evaluate '{"expression":"document.title"}' --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot cdp DOM.getDocument '{"depth":1}' --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot cdp Runtime.evaluate '{"expression":"document.title"}' --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot cdp DOM.getDocument '{"depth":1}' --tab-id <TAB_ID>
 ```
 
 CDP inspection is not a first-choice operation path. If CDP changes state, it becomes Tier 7 fallback and must follow `fallback-operations.md#raw-cdp-fallback`.
@@ -89,7 +91,7 @@ CDP inspection is not a first-choice operation path. If CDP changes state, it be
 Before reporting a browser issue, collect the smallest evidence set that explains it:
 
 - Runtime: `doctor`, `tabs`, `visibility status`.
-- Page state: `get`, `is`, `scan`, or `snapshot`.
+- Page state: `get`, `is`, `snapshot`, or `wait`.
 - Visual proof: `screenshot --annotate` only if visual state matters.
 - Browser diagnostics: `console errors` and `network summary`.
 - Reproduction artifact: `trace` or `record` when the flow is unstable.

package/bin/omnibot-windows-x64/omnibot/skills/omnibot/references/fallback-operations.md

@@ -106,7 +106,8 @@ Use `execute-js` only when:
 - CSS selector operation fails.
 - DOM fallback fails.
 - A complex frontend event must be triggered.
-- Runtime state cannot be read through `read`, `scan`, `get`, or `is`.
+- A CSS-only dropdown or hover menu requires an atomic open-then-select sequence, and `snapshot -i` confirms the option nodes are absent from the accessibility tree.
+- Runtime state cannot be read through `read`, `snapshot`, `get`, or `is`.
 
 Read-only JavaScript fallback:
 
@@ -129,6 +130,19 @@ OMNIBOT_SESSION_TOKEN=repair omnibot execute-js "document.querySelector('button[
 OMNIBOT_SESSION_TOKEN=repair omnibot wait --url "/dashboard" --tab-id <TAB_ID>
 ```
 
+### CSS-only dropdown JavaScript fallback
+
+Use this only after the trigger is known and higher tiers failed: semantic `find` clicked the trigger or found the text, `snapshot -i` could not expose option refs, selector/DOM/mouse could not complete the two-step interaction, and verification showed the selection did not change.
+
+Use a short `execute-js --file` script and verify the user-visible result, not just `{ ok: true }`: selected label, sorted result order, result count, URL/query state, or another page-specific condition.
+
+Detailed recipe: `css-only-dropdown-js.md`.
+
+```bash
+OMNIBOT_SESSION_TOKEN=repair omnibot execute-js --file /tmp/omnibot-css-dropdown.js --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=repair omnibot snapshot -i --tab-id <TAB_ID>
+```
+
 Prefer `--file` for longer scripts:
 
 ```bash
@@ -143,15 +157,15 @@ OMNIBOT_SESSION_TOKEN=repair omnibot snapshot -i --tab-id <TAB_ID>
 Raw CDP is the last resort. Use it only when standard commands and JavaScript fallback cannot complete or inspect the task.
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot cdp Runtime.evaluate '{"expression":"document.title"}' --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot cdp DOM.getDocument '{"depth":1}' --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot cdp Runtime.evaluate '{"expression":"document.title"}' --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot cdp DOM.getDocument '{"depth":1}' --tab-id <TAB_ID>
 ```
 
 After CDP fallback, verify using normal Omnibot reads whenever possible:
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug omnibot get title --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=debug omnibot snapshot -i --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot get title --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=debug-checkout omnibot snapshot -i --tab-id <TAB_ID>
 ```
 
 CDP can bypass application-level event semantics. Treat it as expert-only infrastructure access.

package/bin/omnibot-windows-x64/omnibot/skills/omnibot/references/operation-patterns.md

@@ -8,7 +8,7 @@ Use this file to choose behavior by task. Every page-state workflow must set `OM
 
 Use Read when the agent needs clean rendered page content: article text, search results, feed items, long pages, lazy-loaded text, or a human-readable summary source.
 
-Use targeted state commands when the agent needs one value, visibility, enabled/checked state, layout box, or computed styles. Use scan when the agent needs a structured current-viewport observation for deciding the next action.
+Use targeted state commands when the agent needs one value, visibility, enabled/checked state, layout box, or computed styles. Use `snapshot -i` when the agent needs current visible UI structure for deciding the next action.
 
 ### Preferred sequence
 
@@ -17,9 +17,8 @@ Use targeted state commands when the agent needs one value, visibility, enabled/
 3. `read <URL>` when opening a temporary tab only for reading that URL.
 4. `get title/url/text/html/value/attr/count/box/styles` for narrow evidence.
 5. `is visible/enabled/checked` for boolean state.
-6. `scan --json` only when the next step depends on current viewport structure.
-7. `snapshot -i` when actionable refs are needed.
-8. `dom visible` or `execute-js` fallback only after standard reads cannot access the state.
+5. `snapshot -i` when actionable refs are needed.
+6. `dom visible` or `execute-js` fallback only after standard reads cannot access the state.
 
 ### Example
 
@@ -40,7 +39,7 @@ OMNIBOT_SESSION_TOKEN=research omnibot get count ".result" --tab-id <TAB_ID>
 
 ### Fallback entry point
 
-If `read`, targeted reads, or scan omit needed runtime state, go to `fallback-operations.md#fallback-tier-model`. Start with DOM fallback before JavaScript.
+If `read` or targeted reads omit needed runtime state, go to `fallback-operations.md#fallback-tier-model`. Start with DOM fallback before JavaScript.
 
 ## Click
 
@@ -161,9 +160,23 @@ OMNIBOT_SESSION_TOKEN=form omnibot is checked "#agree" --tab-id <TAB_ID>
 
 For checkboxes, verify boolean state. For selects, verify value or dependent page state.
 
+### CSS-only dropdowns and custom menus
+
+Some dropdowns, hover menus, tooltips, and custom selects are CSS-only widgets. The visible trigger may be accessible, while the option nodes are hidden `<div>` elements without `role`, accessible name, or stable refs.
+
+Use this evidence path before JavaScript:
+
+1. Try semantic `find` or a normal `select`/`click` against the trigger.
+2. Verify whether the selected label, result order, or dependent state changed.
+3. Run `snapshot -i` after opening the control.
+4. If the trigger clicked but options are absent from the accessibility tree, do not keep retrying `find` or stale refs.
+5. Escalate to `fallback-operations.md#css-only-dropdown-javascript-fallback` for an atomic open-then-select script.
+
+`snapshot -i` not showing an option does not prove the DOM node is missing. It may only prove the widget is not exposed through the accessibility tree.
+
 ### Fallback entry point
 
-If native select/check fails, use `fallback-operations.md#selector-fallback`, then JavaScript only if events must be synthesized.
+If native select/check fails, use `fallback-operations.md#selector-fallback`. Use JavaScript only if events must be synthesized or if evidence shows a CSS-only multi-step widget cannot be completed through semantic, snapshot, selector, DOM, or mouse tiers.
 
 ## Navigation
 
@@ -201,9 +214,9 @@ Other navigation commands:
 
 ```bash
 OMNIBOT_SESSION_TOKEN=research omnibot navigate https://example.com
-OMNIBOT_SESSION_TOKEN=research omnibot navigate --same-tab https://example.com
+OMNIBOT_SESSION_TOKEN=research omnibot navigate https://example.com --same-tab --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=research omnibot open https://example.com
-OMNIBOT_SESSION_TOKEN=research omnibot goto https://example.com
+OMNIBOT_SESSION_TOKEN=research omnibot goto https://example.com --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=research omnibot pushstate /dashboard --tab-id <TAB_ID>
 OMNIBOT_SESSION_TOKEN=research omnibot close <TAB_ID>
 OMNIBOT_SESSION_TOKEN=research omnibot tab list
@@ -261,7 +274,7 @@ OMNIBOT_SESSION_TOKEN=checkout omnibot wait --fn "window.appReady === true" --ta
 
 ### Verification
 
-After waiting, read the final condition with `get`, `is`, `scan`, or `snapshot`.
+After waiting, read the final condition with `get`, `is`, `wait`, or `snapshot`.
 
 ### Fallback entry point
 
@@ -277,7 +290,7 @@ Use Extraction for retrieving page text, HTML fragments, links, counts, assets,
 
 1. `get text/html/attr/count` for targeted extraction.
 2. `read --screens N` for clean page text, long pages, or lazy-loaded content.
-3. `scan --json` for structured viewport extraction before action planning.
+3. `snapshot -i` for structured page observation before action planning.
 4. `assets list/export` for resources.
 5. `clipboard read` only when clipboard content is part of the task.
 
@@ -302,31 +315,30 @@ Verify extraction completeness with expected markers in `read`, element counts,
 
 ### Fallback entry point
 
-If content exists only in runtime objects, use `fallback-operations.md#javascript-fallback` after explaining why `read`, `scan`, and `get` could not access it.
+If content exists only in runtime objects, use `fallback-operations.md#javascript-fallback` after explaining why `read`, `snapshot`, and `get` could not access it.
 
 ## Upload
 
 ### When to use
 
-Use Upload when a browser form needs a local file attached through a file input.
+Native upload is currently unavailable in the extension transport. Do not use `omnibot upload` as an agent workflow step until file chooser support is implemented.
 
 ### Preferred sequence
 
-1. Observe the file input or upload control.
-2. Use `upload` against the input selector.
-3. Verify that the uploaded file appeared or form state changed.
+1. Observe the upload UI and confirm the task requires a local file.
+2. Report that native upload is currently unavailable.
+3. If the user still wants a workaround, ask for approval before using page-specific JavaScript or manual browser interaction.
 
 ### Example
 
 ```bash
 OMNIBOT_SESSION_TOKEN=form omnibot snapshot -i --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=form omnibot upload "input[type=file]" /path/to/file.png --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=form omnibot wait --text "file.png" --tab-id <TAB_ID>
+# upload is currently unavailable; do not run omnibot upload here.
 ```
 
 ### Verification
 
-Use visible filename, count, enabled submit state, or application-specific confirmation:
+If upload support is implemented later, verify with visible filename, count, enabled submit state, or application-specific confirmation:
 
 ```bash
 OMNIBOT_SESSION_TOKEN=form omnibot is enabled "button[type=submit]" --tab-id <TAB_ID>
@@ -334,13 +346,13 @@ OMNIBOT_SESSION_TOKEN=form omnibot is enabled "button[type=submit]" --tab-id <TA
 
 ### Fallback entry point
 
-If upload input is hidden behind custom UI, click the semantic upload control first. Use JavaScript only after standard upload cannot reach the input.
+Do not treat JavaScript as a normal upload fallback. File inputs and browser file chooser behavior are security-sensitive and page-specific.
 
 ## Batch
 
 ### When to use
 
-Use Batch for short, known-safe command chains where the target tab and verification are already explicit.
+Use Batch for short, known-safe extension/CDP command chains where the target tab and verification are already explicit. Batch is not a wrapper for normal CLI commands such as `snapshot` or `click`.
 
 ### Preferred sequence
 
@@ -352,8 +364,8 @@ Use Batch for short, known-safe command chains where the target tab and verifica
 ### Example
 
 ```bash
-OMNIBOT_SESSION_TOKEN=research omnibot batch '[{"cmd":"scan"},{"cmd":"snapshot","interactive":true}]' --tab-id <TAB_ID>
-OMNIBOT_SESSION_TOKEN=research omnibot snapshot -i --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=research omnibot batch '[{"cmd":"cdp","method":"Runtime.evaluate","params":{"expression":"document.title","returnByValue":true}}]' --tab-id <TAB_ID>
+OMNIBOT_SESSION_TOKEN=research omnibot get title --tab-id <TAB_ID>
 ```
 
 File example:

package/bin/omnibot-windows-x64/omnibot/skills/omnibot/references/session-and-tabs.md

@@ -1,17 +1,17 @@
-# Sessions and Tabs
+# Workflow Context and Tabs
 
-Sessions and tabs are core reliability concepts, not advanced usage.
+Workflow context tokens and tab ids are core reliability concepts, not advanced usage.
 
-## Session First
+## Context First
 
-`OMNIBOT_SESSION_TOKEN` is mandatory for every workflow.
+`OMNIBOT_SESSION_TOKEN` is mandatory for every workflow. It is a workflow/context token, not a browser session, login session, or tab target.
 
 ```bash
 OMNIBOT_SESSION_TOKEN=research omnibot tabs
 OMNIBOT_SESSION_TOKEN=research omnibot snapshot -i --tab-id <TAB_ID>
 ```
 
-The same token keeps one workflow isolated. Different tokens prevent independent agents from sharing default state.
+The same token keeps one workflow isolated. Different tokens prevent independent agents from sharing refs, trace, recording, temporary resources, and other workflow state.
 
 ## Tab Explicit
 
@@ -41,6 +41,7 @@ Safe sequence:
 
 ```bash
 OMNIBOT_SESSION_TOKEN=research omnibot navigate https://example.com
+# Save the returned full tab id exactly, for example edge-client:123.
 OMNIBOT_SESSION_TOKEN=research omnibot snapshot -i --tab-id edge-client:123
 OMNIBOT_SESSION_TOKEN=research omnibot click --tab-id edge-client:123 @e4
 OMNIBOT_SESSION_TOKEN=research omnibot wait --text "Saved" --tab-id edge-client:123
@@ -68,10 +69,10 @@ Browser integration tests must only close tabs they created:
 
 ## Why Both
 
-- Session = workflow isolation.
+- Workflow context token = agent/workflow state isolation.
 - Tab ID = page targeting.
 
-Session alone does not prove which page receives a command. Tab ID alone does not isolate concurrent agents. Use both.
+The workflow token alone does not prove which page receives a command. Tab ID alone does not isolate concurrent agents. Use both.
 
 ## Ref Scope
 
@@ -102,7 +103,7 @@ OMNIBOT_SESSION_TOKEN=checkout
 Agent C:
 
 ```bash
-OMNIBOT_SESSION_TOKEN=debug
+OMNIBOT_SESSION_TOKEN=debug-checkout
 ```
 
 Each agent must:
@@ -143,13 +144,13 @@ Claiming does not replace `--tab-id`. It only documents or coordinates ownership
 
 ## visibility
 
-Visibility controls where automation runs:
+Visibility controls where automation is intended to run:
 
 - `visible`: controls the user's already-open real browser tabs and preserves visible session state.
 - `background`: avoids foregrounding tabs unless requested.
-- `dedicated-profile`: launches or uses a separate profile.
-- `headless`: uses a headless browser context.
-- `launch`: starts a dedicated-profile or headless browser with an explicit user data dir.
+- `dedicated-profile`: planned separate-profile mode.
+- `headless`: planned headless browser context.
+- `launch`: currently reports planned configuration only; it does not launch a browser.
 
 Examples:
 
@@ -157,8 +158,7 @@ Examples:
 omnibot visibility status
 omnibot visibility set visible
 omnibot visibility set background
-omnibot visibility set headless --user-data-dir /tmp/omnibot-headless
 omnibot visibility launch headless --user-data-dir /tmp/omnibot-headless
 ```
 
-Headless and dedicated-profile modes do not automatically share the user's current tabs or login state. Verify login state before assuming access.
+Headless and dedicated-profile modes do not automatically share the user's current tabs or login state. Because `visibility launch` is currently planned-only, verify actual browser availability with `doctor` and `tabs` before assuming access.

package/package.json

@@ -1,29 +1,29 @@
-{
-  "name": "@omniaibot/win-x64",
-  "version": "1.1.2",
-  "description": "omnibot executable for Windows x64",
-  "os": [
-    "win32"
-  ],
-  "cpu": [
-    "x64"
-  ],
-  "bin": {
-    "omnibot": "bin/omnibot-windows-x64/omnibot-windows-x64.exe"
-  },
-  "files": [
-    "bin/"
-  ],
-  "keywords": [
-    "mcp",
-    "browser",
-    "automation",
-    "omnibot"
-  ],
-  "author": "Unagi-cq",
-  "license": "MIT",
-  "repository": {
-    "type": "git",
-    "url": "git+https://github.com/DennisJcy/omnibot.git"
-  }
-}
+{
+  "name": "@omniaibot/win-x64",
+  "version": "1.3.0",
+  "description": "omnibot executable for Windows x64",
+  "os": [
+    "win32"
+  ],
+  "cpu": [
+    "x64"
+  ],
+  "bin": {
+    "omnibot": "bin/omnibot-windows-x64/omnibot-windows-x64.exe"
+  },
+  "files": [
+    "bin/"
+  ],
+  "keywords": [
+    "browser",
+    "automation",
+    "agent",
+    "omnibot"
+  ],
+  "author": "Unagi-cq",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/DennisJcy/omnibot.git"
+  }
+}