How to Use the Agent-Reach Configure Command with Browser Cookie Extraction

TLDR: Run agent-reach configure --from-browser <browser> to automatically extract authentication cookies from your local Chrome, Firefox, Edge, or Safari browser and populate your ~/.agent-reach/config.yaml with credentials for Twitter/X, XiaoHongShu, Bilibili, and Xueqiu.

The Agent-Reach CLI provides a configure sub-command that eliminates manual cookie copying by reading authentication tokens directly from your browser's cookie store. This automation supports multiple social platforms and handles the complex mapping of browser cookies to configuration keys required by the tool.

Basic Command Syntax

The browser extraction flag attaches to the standard configure command:

agent-reach configure --from-browser <BROWSER>

Supported browser values include chrome, firefox, edge, and safari. The command requires the browser to be installed on your local machine and accessible in standard profile directories.

How Browser Cookie Extraction Works

The extraction pipeline implemented in agent_reach/cookie_extract.py follows a four-stage process when triggered from the CLI entry point.

Step 1: CLI Argument Parsing and Dispatch

In agent_reach/cli.py (lines 81-91), the configure sub-parser defines the --from-browser option. When invoked, the _cmd_configure handler (lines 26-34) detects this flag and immediately routes execution to configure_from_browser instead of the interactive configuration mode.

Step 2: Cookie Extraction Backend Selection

The extract_all(browser) function in agent_reach/cookie_extract.py (lines 53-66) attempts to import rookiepy (a Rust-based extractor) as the preferred engine, falling back to browser-cookie3 if unavailable. This backend opens the browser’s SQLite cookie store and iterates over all stored cookies, matching them against the PLATFORM_SPECS table defined in lines 13-39.

The function returns a dictionary mapping platform keys (twitter, xhs, bilibili, xueqiu) to their respective extracted cookie data (lines 42-52).

Step 3: Platform-Specific Cookie Mapping

configure_from_browser processes the extraction results and transforms browser cookie names into Agent-Reach configuration keys:

• Twitter/X: Extracts auth_token and ct0, storing them as twitter_auth_token and twitter_ct0 (lines 50-58). The function also synchronizes these values to the legacy xfetch session file and the bird environment file.
• XiaoHongShu: Saves the complete cookie string as xhs_cookie (lines 66-72).
• Bilibili: Maps SESSDATA to bilibili_sessdata and optional bili_jct to bilibili_csrf (lines 73-80).
• Xueqiu: Persists cookies only if the required xq_a_token is present, saving it as xueqiu_token (lines 85-92).

Step 4: Configuration Persistence

All extracted values are written to the user-level configuration file (~/.agent-reach/config.yaml) via the Config.set method from agent_reach/config.py. The utility uses make_private_dir from agent_reach/utils/paths.py to ensure secure file permissions during creation.

Practical Examples

Extract from Chrome (the default choice for most desktop environments):

agent-reach configure --from-browser chrome

Extract from Firefox on Linux systems where it is the default browser:

agent-reach configure --from-browser firefox

Successful execution produces formatted output indicating which platforms were configured:

Extracting cookies from chrome...

  ✅ Twitter/X: auth_token + ct0
  ✅ XiaoHongShu: 12 cookies
  ✅ Bilibili: SESSDATA + bili_jct
  ✅ Xueqiu: 7 cookies (含 xq_a_token)

✅ Cookies configured! Run `agent-reach doctor` to see updated status.

Troubleshooting Browser Access Issues

If the browser is currently running or the cookie database is locked, the extraction will fail silently for that specific browser. The console output displays a warning indicator:

  -- Could not read browser cookies (browser might be open or password was denied)

To resolve this, close the browser completely before running the command. Ensure your user account has read permissions to the browser's profile directory (typically ~/Library/Application Support/Google/Chrome on macOS or ~/.config/google-chrome on Linux).

Summary

• The agent-reach configure --from-browser <browser> command automates cookie extraction from local browser stores.
• The implementation in agent_reach/cookie_extract.py uses rookiepy or browser-cookie3 to read SQLite cookie databases.
• Supports four platforms: Twitter/X, XiaoHongShu, Bilibili, and Xueqiu, with specific mappings to configuration keys.
• Credentials are persisted to ~/.agent-reach/config.yaml via the Config.set method.
• Always run agent-reach doctor after configuration to verify platform connectivity.

Frequently Asked Questions

Which browsers are supported for cookie extraction?

Agent-Reach supports Chrome, Firefox, Edge, and Safari. The underlying extraction logic in agent_reach/cookie_extract.py attempts to locate the standard profile directories for each browser, though availability depends on your operating system and installation method.

What happens if my browser is running when I execute the command?

If the browser process is active, the cookie database file remains locked, preventing read access. The CLI will display a warning message (-- Could not read browser cookies...) and skip that browser. You must close the browser completely before extraction.

How are the extracted cookies stored on my system?

Cookies are written to the YAML configuration file at ~/.agent-reach/config.yaml using the Config.set method. Twitter/X cookies receive special handling—they are simultaneously synced to supplemental files including the xfetch session store and bird environment configuration to maintain compatibility with legacy workflows.

Do I need to install additional Python packages for this to work?

The system attempts to use rookiepy first (a Rust-based extractor), but falls back to browser-cookie3 if unavailable. If neither package is installed, the extraction will fail. Most standard installations include one of these dependencies, but you can manually install browser-cookie3 via pip if you encounter import errors.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}