feishu-sheet-tabs

Create and organize multiple tabs (worksheets/pages) inside an existing Feishu sheet when the normal feishu_sheet API cannot create new worksheet tabs. Use when the user wants one spreadsheet split into categories/pages/tabs, asks for pagination inside a sheet, or wants separate worksheet tabs such as 总览 / Skills / Workflows / Templates / Content. Prefer this skill only after confirming feishu_sheet API lacks tab-creation support.

Use this skill to add worksheet tabs inside an existing Feishu sheet by browser automation when feishu_sheet API can create/read/write spreadsheets but cannot create worksheet tabs.

Core conclusion

Current feishu_sheet API supports spreadsheet-level actions like:

• create
• info
• read
• write
• append
• find
• export

It can target an existing sheet via sheet_id, but it does not expose a direct action for creating worksheet tabs/pages.

When the user explicitly wants tabs/pages inside one spreadsheet, switch to browser automation.

Preconditions

Before using browser automation, ensure all of the following:

1. The target spreadsheet URL/token is known.
2. The user is logged into Feishu in a browser tab.
3. Prefer using the user's real Chrome tab via Browser Relay when available.
4. If Browser Relay is not attached, isolated browser may work only if it has a valid Feishu login state.

Preferred operating mode

Mode A: User's Chrome tab (preferred)

Use browser with profile="chrome" if the user has attached the tab via OpenClaw Browser Relay.

Why:

• stable login state
• real user session
• fewer auth surprises

Mode B: OpenClaw managed browser

Use profile="openclaw" only if it is already logged into Feishu.

Reliable workflow

Step 1: Verify API limitation first

Do not claim “can’t do it” without checking.

Confirm from tool docs / current tool signature that feishu_sheet has no add_sheet / create_worksheet / add_tab action.

Step 2: Open the spreadsheet in browser

Use browser automation to open the spreadsheet.

Step 3: Inspect runtime objects

Feishu Sheets exposes internal JS objects on the page. In practice, these were found useful:

• window.spread
• window.spreadApp

Relevant methods discovered from runtime introspection:

• spread.addSheet(name)
• spread.renameSheet(sheetId, name)
• spread.copySheet(...)
• spread.moveSheet(...)
• spread.hideSheet(...)

Step 4: Read current sheet ids/names

Use page evaluation to inspect current sheets before mutation.

Pattern:

js

window.spread.sheets.map((s,i)=>({
  i,
  id: s.id?.() ?? s._id ?? s.sheetId ?? null,
  name: s.name?.() ?? s._name ?? null
}))

Step 5: Rename default first tab if needed

Typical pattern:

• rename Sheet1 → 总览

Step 6: Add tabs with spread.addSheet(name)

For example:

• Skills
• Workflows
• Templates
• Content

Step 7: Fill each tab with feishu_sheet write

After tabs exist and their sheet_ids are known, return to feishu_sheet for structured writes. This is more stable than trying to type cell-by-cell in browser automation.

Proven pattern from this workspace

For spreadsheet:

• https://bytedance.larkoffice.com/sheets/Bf6qsMV9fhqrD6tPE6TcQhF7nEe

The following sequence worked:

1. Inspect window.spread
2. Discover spread.addSheet and spread.renameSheet
3. Read current sheet list and ids
4. Rename Sheet1 to 总览
5. Create:
   • Skills
   • Workflows
   • Templates
   • Content
6. Use feishu_sheet write with returned sheet ids to populate each tab

Example evaluation snippets

Discover methods

js

function protoMethods(obj,name){
  if(!obj) return [];
  const out=[];
  let p=Object.getPrototypeOf(obj);
  let depth=0;
  while(p && depth<4){
    for(const k of Object.getOwnPropertyNames(p)){
      if(k==='constructor') continue;
      if(/sheet|tab|workbook|insert|create|add|page|name|rename/i.test(k)) out.push(`${name}.${k}`);
    }
    p=Object.getPrototypeOf(p); depth++;
  }
  return [...new Set(out)];
}

Rename + create tabs

js

async () => {
  const spread = window.spread;
  const results = [];

  const current = spread.sheets.map(s => ({
    id: s.id?.() ?? s._id ?? null,
    name: s.name?.() ?? s._name ?? null
  }));

  const first = current[0];
  if (first?.name === 'Sheet1') {
    results.push(await spread.renameSheet(first.id, '总览'));
  }

  for (const name of ['Skills','Workflows','Templates','Content']) {
    const names = spread.sheets.map(s => s.name?.() ?? s._name);
    if (!names.includes(name)) {
      results.push(await spread.addSheet(name));
    }
  }

  return spread.sheets.map((s,i)=>({
    i,
    id: s.id?.() ?? s._id ?? null,
    name: s.name?.() ?? s._name ?? null
  }));
}

Important caveats

1. Prefer internal methods over brittle UI clicking Clicking + in the UI was less reliable than runtime JS calls.

2. Do not assume a visible plus button is the right entry In this case, one nearby button turned out to be a template entry, not “new tab”.

3. Use browser only for the tab creation step Once tabs are created, switch back to feishu_sheet write.

4. Inspect current ids before writing New tabs get generated sheet_ids such as GxGIGa, 9dJYiB, etc. Use actual returned ids.

5. Avoid blind UI automation first Inspect runtime objects and methods before pixel-clicking.

When to use this skill

Use this skill when the user says things like:

• “把这个表做成分页”
• “按类别分开 sheet”
• “给这张飞书表加几个页签”
• “我就想分页，你想办法解决”

Output expectation

When done, report:

• which tabs were created/renamed
• which tab ids were found
• whether data filling was also completed
• whether browser login / relay was required