- Auto-click New Conversation if session has only 1 message
- Map Anthropic models (claude-3-7-sonnet) to Antigravity UI models
- Refactor waitForReply to check for Cancel/Stop button presence to
detect generation completion reliably, with text stability fallback
- Replace document.execCommand (deprecated) with CDP Input.insertText
- Use Input.dispatchMouseEvent to physically click + focus the Lexical editor
before text injection (fixes focus issues with JS-only .focus())
- Improve getLastAssistantReply: strip echoed user message, thinking blocks,
Copy button text, and de-duplicate repeated content artifacts
- New command: opencli antigravity serve --port 8082
- Starts HTTP server compatible with Anthropic /v1/messages API
- Connects to Antigravity via CDP (OPENCLI_CDP_ENDPOINT)
- Uses Input.dispatchKeyEvent for reliable Enter key submission
- Polls for reply with text-change detection + 3s stability check
- Precise DOM walker for extracting last assistant reply
- Lazy CDP connection (connects on first request)
- Auto-reconnect on CDP connection loss
- CORS headers for Claude Code compatibility
Usage:
OPENCLI_CDP_ENDPOINT=http://127.0.0.1:9224 opencli antigravity serve
ANTHROPIC_BASE_URL=http://localhost:8082 claude
formatPostTime() used local timezone methods, causing test failure
on UTC CI servers. XHS API timestamps are Beijing time (UTC+8),
so use explicit UTC offset with getUTC*() methods.
- wikiFetch return Promise<unknown> instead of Promise<any>
- Add WikiSearchResult type, remove r: any
- Type wikiFetch responses with inline type assertions
- Only append ... to abstract when actually truncated
- Remove fetchCreatorNotesByCdp() and captureNoteDetailApiPayload() raw
WebSocket code (~240 lines) — adapters should use IPage, not raw CDP
- Replace direct CDP WebSocket with IPage.evaluate() in-page fetch
- Fix page: any → IPage in all function signatures
- Simplify to two-tier fallback: API+interceptor → DOM parse
- Rebase onto latest main (resolves cdp.ts/daemon.ts conflicts)
The build manifest includes antigravity/serve which collides with the
hardcoded antigravity serve in cli.ts. Add a guard to skip registry
entries whose subcommand already exists in the site group.
# Please enter a commit message to explain why this merge is necessary,
# especially if it merges an updated upstream into a topic branch.
#
# Lines starting with '#' will be ignored, and an empty message aborts
# the commit.
Replace eager close-window (which caused race conditions when
parallel commands shared the window) with an idle-based timer:
- Window auto-closes 30s after the last command completes
- Each incoming command resets the idle timer
- Consecutive commands reuse the same window (faster)
- No race conditions with parallel execution
- Close-window action kept for explicit cleanup if needed
- Add 'close-window' action to extension protocol and background.ts
- Add Page.closeWindow() method to send close-window command
- browserSession() now closes automation window in cleanup
- Remove domain pre-navigation + 2s wait from main.ts (CDP handles
cross-domain cookies natively, no same-origin workaround needed)
- Net effect: commands run faster, no stale windows left behind
- Delete unused extension/src/executor.ts (chrome.scripting experiment)
- Remove 15 no-op backward-compat exports from doctor.ts
- Remove getTokenFingerprint no-op from browser/index.ts
- Rename PlaywrightMCP → BrowserBridge across all source files
(backward-compat alias kept in mcp.ts and browser/index.ts)
- Remove unnecessary host_permissions from extension manifest
- Sync extension package.json version to 0.2.0
- All 14 tests pass
All opencli operations now run in a dedicated Chrome window instead
of hijacking the user's active tab. The automation window:
- Created on first command via chrome.windows.create({ focused: false })
- 1280x900 viewport, auto-cleaned up when closed
- All tabs resolved within this window only
- User's main browsing session is never touched
Tested: twitter trending ✅, zhihu hot ✅
Both commands now scroll the conversation list to load more items
before processing. Scrolls up to 20-30 times, stops after 3
consecutive scrolls with no new items loaded.
Previously limited to ~14 visible conversations, now loads as many
as needed (up to --max).
- Rewrite accept.ts: use [data-testid=conversation] click-based approach
instead of extracting href links (requests page has no /messages/xxx links)
- Support comma-separated keywords for OR matching (e.g. '群,微信')
- Add timeoutSeconds: 600 (10 min) for batch DM operations
- Bump default OPENCLI_BROWSER_COMMAND_TIMEOUT from 45s to 60s
- Track visited conversations to avoid infinite loops
Usage:
opencli twitter accept --keyword '微信' --max 20
Workflow:
1. Navigate to /messages/requests
2. Click into each conversation
3. If message contains keyword, click Accept
4. After accept (auto-redirects to /messages), go back to requests
5. Repeat until --max reached or no more matches
process.execPath is always plain 'node' even under tsx,
so .ts files could not be executed. Use --import tsx/esm
flag to enable TypeScript loading in spawned daemon.
Exponential backoff:
- Reconnect delay: 2s, 4s, 8s, 16s, ..., capped at 60s
- Resets to base delay on successful connection
- Reduces idle CPU waste vs fixed 3s reconnect
Screenshot via CDP Page.captureScreenshot:
- New 'screenshot' action in protocol (5th action)
- Supports format (png/jpeg), quality, fullPage
- Full-page: uses Emulation.setDeviceMetricsOverride for scroll height
- CLI-side: page.screenshot() with optional file save
- Extension build: 9.81KB (+1.7KB from 8.11KB)
Inspired by bb-browser's architecture patterns.
Feishu uses custom 'Lark Framework' (Chromium-based but NOT Electron).
CDP port test failed — --remote-debugging-port has no effect.
Uses AppleScript + clipboard approach (same as WeChat/ChatGPT).
Commands: status, send, read, search (Cmd+K), new (Cmd+N)
Includes adapter READMEs (EN+ZH).
- Remove feishu and wechat adapters (not tested yet, will re-add later)
- Remove their rows from README.md and README.zh-CN.md
- Significantly polish CLI-ELECTRON.md skill guide:
- Add Electron detection guide (check for Electron Framework)
- Add Non-Electron AppleScript pattern section
- Add port assignment table for all CDP adapters
- Improve code examples with real working TypeScript
WeChat Mac is native Cocoa (not Electron), so CDP is not available.
Uses AppleScript + clipboard automation instead:
- status: check if WeChat is running
- send: paste + Enter in active conversation
- new: Cmd+N for new chat
- search: Cmd+F and type query
- read: Cmd+A → Cmd+C to copy chat content
Total: 30 sites · 156 commands
- Remove --remote-allow-origins from antigravity README, README.zh-CN, SKILL.md (not needed for local usage)
- Update ChatGPT README to document both AppleScript and CDP approaches
- Document ChatGPT Electron launch: /Applications/ChatGPT.app/Contents/MacOS/ChatGPT --remote-debugging-port=9224
* feat(browser): add CDP remote connection support for server environments
This feature enables OpenCLI to connect to a Chrome browser running on a
different machine (e.g., your local computer) from a headless server
environment via Chrome DevTools Protocol (CDP).
Server environments (CI, cloud VMs, headless Linux) cannot run Chrome with
a GUI or install the Playwright MCP Bridge extension. This makes it
impossible to use OpenCLI commands that require browser authentication.
Add support for the `OPENCLI_CDP_ENDPOINT` environment variable, which
tells OpenCLI to connect to a remote Chrome instance via CDP instead of
using the local extension mode.
1. Start Chrome with remote debugging on local machine:
```
chrome --remote-debugging-port=9222 --user-data-dir="$HOME/chrome-debug"
```
2. Create SSH tunnel to forward port to server:
```
ssh -R 9222:localhost:9222 your-server
```
3. Run OpenCLI on server:
```
export OPENCLI_CDP_ENDPOINT="http://localhost:9222"
opencli bilibili hot --limit 5
```
- src/browser.ts: Add CDP endpoint detection in buildMcpArgs()
- src/doctor.ts: Show CDP mode status in doctor report
- README.md: Add "Remote Chrome (Server/Headless)" section
- README.zh-CN.md: Add corresponding Chinese documentation
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
* docs: extract CDP connection guide into separate files
* docs: clarify CDP vs SSH/Proxy distinction in CDP guides
* docs: restructure CDP guides into 3 distinct phases (preparation, tunnel, execution)
---------
Co-authored-by: ByteYue <yj976240184@gmail.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
jackwener