功能描述

Automates DataHive sign-in using a magic link workflow: requests the link, retrieves it from Gmail via gog, and opens it in a Chrome DevTools tab.

使用说明 (SKILL.md)

datahive-installer

Name: DataHive Installer
Author: tuleyko

Automates installation of DataHive node.

DataHive context (for operators)

DataHive is a platform for high-quality AI training datasets and analytics use cases across domains such as e-commerce, reviews, audio, image, and video, with an emphasis on decentralized and passive data collection workflows. This skill is limited to dashboard authentication automation, specifically requesting a magic link, retrieving it, and opening the login tab, and does not cover dataset management.

Trigger phrases

Use this skill when the user says things like:

"run datahive-installer"
"log me into DataHive"
"start DataHive login flow"
"execute DataHive installer"
"setup DataHive"
"install DataHive"

Execution rule

Always execute this skill in the exact sequence defined in strict order, exactly as written in this SKILL.md. Do not skip, reorder, or parallelize steps unless the user explicitly asks to change the sequence.

What this skill does

Detect platform (ubuntu or macos) and stop with an error for any other OS.
Install runtime prerequisites for the detected platform.
Launch Chrome in persistent background mode with CDP enabled.
Read your authenticated gog account email.
Request a DataHive magic link via API.
Find the latest DataHive login email in Gmail.
Extract the magic login URL.
Open the magic URL in a new browser tab over CDP.

Step 0 — Detect platform

Run:

./scripts/0_detect_platform.sh

Expected output:

ubuntu for Ubuntu hosts
macos for macOS hosts

Any other value or non-zero exit means unsupported platform.

Step 1 — Install prerequisites (platform-aware)

Run:

PLATFORM="$(./scripts/0_detect_platform.sh)" ./scripts/1_install_prerequisites.sh

Behavior by platform:

ubuntu: installs Chrome + xvfb via apt, applies managed extension policy, installs websocat.
macos: installs Chrome via Homebrew cask (if missing), applies managed extension policy in /Library/Managed Preferences/com.google.Chrome.plist, installs websocat.

Step 2 — Launch browser in persistent background mode (platform-aware)

Run:

PLATFORM="$(./scripts/0_detect_platform.sh)" ./scripts/2_launch_chrome_supervisor.sh

Behavior by platform:

ubuntu: launches google-chrome under xvfb-run.
macos: launches /Applications/Google Chrome.app/Contents/MacOS/Google Chrome directly.

Optional checks:

curl -sf http://localhost:9222/json/version

tail -f "$HOME/.chrome-datahive/chrome.log"

Step 3 — Get email from gog

gog auth list --json

Use the default account email from output (example: [email protected]).

Step 4 — Request magic link

Use the helper script with your gog email:

./scripts/4_request_magic_link.sh \x3CEMAIL>

(or EMAIL=\x3CEMAIL> ./scripts/4_request_magic_link.sh)

Expected response:

{"message":"If this email exists, a login link has been sent"}

Step 5 — Find latest DataHive magic-link email

gog gmail messages search 'from:datahive newer_than:7d' --max 10 --account \x3CEMAIL>

Take the top/latest message ID and fetch it:

gog gmail get \x3CMESSAGE_ID> --account \x3CEMAIL> --json

Extract the URL from the HTML body:

https://dashboard.datahive.ai/auth?token=...

Step 6 — Open magic link

Run:

TARGET_URL='https://dashboard.datahive.ai/auth?token=\x3CTOKEN>' ./scripts/6_open_magic_link.sh

Notes

Magic links expire quickly (about 15 minutes).
Treat magic links as secrets; do not share them.
If CDP is not available, start Chrome with remote debugging enabled.

安全使用建议

Before installing or running this skill, consider the following: - It requires sudo and will modify system-managed Chrome policies to force-install a browser extension (system-level change that persists). Ask the author why a system-wide extension is necessary and for the extension's Web Store page or source code to audit it. - The skill reads your Gmail messages via the gog CLI. Ensure gog is properly configured and you understand that the skill will fetch email contents (magic links are sensitive secrets). - The scripts assume Homebrew in places where apt would be expected (Ubuntu), which suggests sloppy or untested behavior — run this only in a sandbox or VM first. - If you need the functionality, prefer a less-privileged approach: run the sequence manually, run scripts in an isolated environment, or ask for a version that doesn't modify system policies and that documents required privileges. - If you proceed: review the extension ID (bonfdkhbkkdoipfojcnimjagphdnfedb) in the Chrome Web Store, inspect all scripts yourself, and avoid running with sudo on a production machine. If you are unsure, decline and request clarification from the publisher or run in a disposable VM/container.

功能分析

Type: OpenClaw Skill Name: datahive-installer Version: 0.1.0 The skill automates the installation of a 'data collection' platform by force-installing a specific browser extension (ID: bonfdkhbkkdoipfojcnimjagphdnfedb) using system-level managed policies on both Ubuntu and macOS, which requires sudo and makes the extension difficult for users to remove. It also programmatically accesses the user's Gmail via the 'gog' utility to search for and extract authentication magic links. While these actions are aligned with the stated purpose of automating DataHive setup, the use of administrative privileges to enforce browser persistence and the automated extraction of credentials from private emails represent significant security and privacy risks (scripts/1_install_prerequisites.sh, scripts/4_request_magic_link.sh, and SKILL.md).

能力评估

⚠ Purpose & Capability

The stated purpose is to request a DataHive magic link from the API, read the email from Gmail (via gog), and open the link in a Chrome DevTools tab. That basic flow is coherent with the required binaries (gog, curl, websocat). However the scripts also force-install a Chrome extension via system-managed policies (writes /etc/opt/chrome/policies/managed/extensions.json on Linux and /Library/Managed Preferences/com.google.Chrome.plist on macOS). Forcing a system-level Chrome extension and editing managed preferences is a high-privilege, persistent action that is not clearly justified by a simple magic-link login flow. Additionally, install_websocat requires Homebrew even on Ubuntu (where the script uses apt for other packages), which is an implementation inconsistency.

⚠ Instruction Scope

SKILL.md directs running multiple scripts that perform privileged actions (sudo apt installs, creation of system policy files, modifying /Library/Managed Preferences). The scripts read your gog authenticated account and Gmail messages (expected for retrieving the link), but they also make persistent system changes (Chrome policy) and run a long‑running Chrome supervisor in background. The instructions assume use of sudo and do not surface that requirement in the metadata. The strict execution-order requirement plus system-wide policy changes grants the skill broad discretion over the user's browser environment beyond what's needed to just open a single link.

⚠ Install Mechanism

There is no registry install spec, but the included scripts perform network installs and system modifications: they add Google's apt repo and install google-chrome via apt on Linux, use Homebrew casks on macOS, and force-install an extension from clients2.google.com. Downloads are from legitimate Google endpoints (chrome repo, clients2.google.com), but the process requires sudo and writes to system policy locations. The scripts' method for installing websocat (via brew) is odd on Ubuntu and could cause failures — a sign of sloppy/unsafe assumptions. Writing system-managed browser policies is a high-risk install behavior.

⚠ Credentials

The skill declares no required environment variables, but it implicitly requires: sudo/root privileges to write system policies and install packages; a configured gog account (so the skill can run gog auth list and gog gmail get); and the ability to start Chrome with a remote-debugging port. Requiring system-wide Chrome policy changes and sudo is disproportionate for a task that could instead open a link in the user's browser or use a temporary browser profile. The skill will read Gmail message contents via gog (which is consistent with the stated goal), but that is sensitive and should be explicitly declared.

⚠ Persistence & Privilege

The skill makes persistent, system-level changes: it writes Chrome managed policies (system directories on Linux and macOS) which force-install an extension and require sudo. It also starts a background Chrome supervisor process that will persist in the user's session and writes files under $HOME/.chrome-datahive. The skill does not use always:true, but the combination of persistent system policy changes plus background supervisor increases blast radius and is a significant privilege escalation relative to just opening a login URL.

版本历史

v0.1.0

Initial release of datahive-installer. - Automates DataHive dashboard login using a magic link workflow. - Detects platform (Ubuntu or macOS) and installs required dependencies. - Launches Chrome in persistent background mode with DevTools enabled. - Retrieves magic login link from Gmail via gog and opens it automatically. - Enforces a strict, stepwise execution sequence. - Limited to dashboard authentication; does not handle dataset management.

元数据

Slug datahive-installer

版本 0.1.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题