chromeflow 0.1.32 → 0.1.33

package/CLAUDE.md

@@ -22,11 +22,11 @@ Do NOT ask "should I open the browser?" — just do it. The user expects seamles
    `scroll_page` then retry, or use `highlight_region` to show the user. Never use
    `osascript`, `applescript`, or any shell command to control the browser.
 
-2. **Never use `take_screenshot` to find element positions or confirm actions.**
-   `get_elements` returns exact DOM coordinates — always use that first. `get_page_text`
-   tells you what happened after an action — always use that before reaching for a screenshot.
-   `take_screenshot` is only for when you genuinely have no idea what the page looks like
-   and DOM queries can't help. It is a last resort, not a routine check.
+2. **Never use `take_screenshot` to read page content.** After `scroll_page`, after
+   `click_element`, after navigation — always call `get_page_text`, not `take_screenshot`.
+   `get_page_text` returns up to 20,000 characters; if truncated it tells you the next
+   `startIndex` to paginate. Screenshots are only for locating an element's pixel position
+   when DOM queries have already failed. Never take more than 1–2 screenshots in a row.
 
 3. **Use `wait_for_selector` to wait for async page changes** (build completion, modals,
    toasts). Never poll with repeated `take_screenshot` calls.

package/dist/tools/capture.js

@@ -60,14 +60,18 @@ After filling, call wait_for_click only if the user needs to review/confirm; oth
     "get_page_text",
     `Get the visible text content of the current page without taking a screenshot.
 Use this instead of take_screenshot whenever you need to read what's on the page \u2014 errors, build status, form labels, confirmation messages, etc.
-Only use take_screenshot when you need to locate an element's pixel position for highlight_region.`,
+Returns up to 20,000 characters at a time. If the response ends with "... (N more characters)", call again with startIndex to read the next chunk.
+Never use take_screenshot just to read page content \u2014 paginate with startIndex instead.`,
     {
       selector: z.string().optional().describe(
         `CSS selector to scope the extraction (e.g. 'main', '.error-toast', '[data-testid="status"]'). Omit to auto-extract from the main content area.`
+      ),
+      startIndex: z.number().optional().describe(
+        "Character offset to start from. Use this to read past the first 20,000 characters \u2014 the response will tell you the next startIndex when more content exists."
       )
     },
-    async ({ selector }) => {
-      const response = await bridge.request({ type: "get_page_text", selector });
+    async ({ selector, startIndex }) => {
+      const response = await bridge.request({ type: "get_page_text", selector, startIndex });
       if (response.type !== "page_text_response") throw new Error("Unexpected response");
       const text = response.text;
       return {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "chromeflow",
-  "version": "0.1.32",
+  "version": "0.1.33",
   "description": "Browser guidance MCP server for Claude Code — highlights, clicks, fills, and captures from the web so you don't have to.",
   "type": "module",
   "bin": {