An OpenClaw skill that gives your agent web browsing, data extraction, content transformation, web research, and browser automation capabilities via the Tabstack API.

Uses @tabstack/sdk v2.

All commands support --geo CC for region-specific content (ISO country code).

The automate command also supports --guardrails (safety constraints) and --data (JSON context for form filling).

• OpenClaw gateway running
• Node.js >= 20 (available inside the OpenClaw container)
• A Tabstack API key

Copy the tabstack/ directory into your OpenClaw workspace:

cp -r tabstack/ ~/.openclaw/workspace/skills/tabstack/
cd ~/.openclaw/workspace/skills/tabstack && npm install

Set your API key:

openclaw config set env.TABSTACK_API_KEY "your-key-here"

Restart the gateway to pick up the new skill.

unzip tabstack.skill -d ~/.openclaw/workspace/skills/
cd ~/.openclaw/workspace/skills/tabstack && npm install
openclaw config set env.TABSTACK_API_KEY "your-key-here"

Validate the skill structure:

make validate

Package for distribution:

make package

This creates tabstack.skill (a zip archive) that can be shared with other OpenClaw users.

1. OpenClaw discovers SKILL.md in the workspace skills directory
2. The skill's description tells the agent when to use it
3. When triggered, the agent reads SKILL.md for instructions
4. The agent runs npx tsx ./tabstack.ts <command> via the exec tool
5. The CLI wrapper calls the Tabstack API and returns results to stdout

tabstack/
├── SKILL.md              — Skill definition and agent instructions
├── tabstack.ts           — CLI wrapper for the Tabstack SDK (v2)
├── package.json          — Dependencies (@tabstack/sdk ^2.2.0, tsx)
└── references/
    └── examples.md       — JSON schema patterns and generate recipes