Browser Use Skill Creator

Browser Use Skill Creator

Create browser automation skills using the Browser Use Cloud API.

Quick Start

Use the skill creation script at apps/api/scripts/create_skill.py:

cd apps/api

# List available templates
uv run python scripts/create_skill.py list

# Create from template and wait for completion
uv run python scripts/create_skill.py create -t youtube-search -w

# Create custom skill
uv run python scripts/create_skill.py create \
  -p "Go to example.com and extract data" \
  -g "Extract data from example.com" \
  -w

# Check status of existing skill
uv run python scripts/create_skill.py status <skill-id>

Available Templates

Creating Custom Skills

When creating a custom skill, provide:

1. Agent Prompt (-p): Step-by-step instructions for the browser agent
2. Goal (-g): Brief description of what the skill accomplishes

Example:

uv run python scripts/create_skill.py create \
  -p "Go to Twitter/X.com, click compose, write the specified message, and post it" \
  -g "Post a tweet to X.com" \
  -w

Skill Lifecycle

1. Create: Skill enters recording state while Browser Use captures the workflow
2. Generate: Skill enters generating state while the API builds the automation
3. Finished: Skill is ready to use with defined parameters and output schema

After Creation

Once a skill is created, add it to the codebase:

1. Note the skill ID from the output
2. Add parameter schema to apps/api/app/skills.py
3. Add formatter and config to apps/api/app/skill_definitions.py
4. Register in register_all_skills()
5. Update apps/api/SKILLS.md with the new skill entry

See existing skills in apps/api/app/skill_definitions.py for reference patterns.

Requirements

• BROWSER_USE_API_KEY must be set in apps/api/.env
• Run from the apps/api directory