SQL Generator (Pro) for Openclaw

1. Ensure the agent environment meets the runtime requirements:

python --version

Python 3.8+ is required.

2. Install the optional database drivers that match your target database:

pip install psycopg2
pip install pymysql
pip install pyodbc
pip install sqlparse

3. Configure database credentials securely through environment variables or a config file. Do not hardcode secrets in source files.

4. Use the skill inside an AI agent platform that supports SKILL.md workflows, such as Claude Code, Cursor, Codex, or Gemini CLI.

5. Initialize schema-aware generation in your application or agent flow:

from sql_gen_tool import ProFeatures

pro = ProFeatures(db_url="database://user:pass@localhost/mydb")
pro.connect_schema()

6. Optionally tune schema refresh and batch generation behavior:

pro.configure(
    schema_refresh="on-demand",
    include_views=True,
    include_indexes=True,
    dialect="database"
)

7. For team workflows, prepare a prompt list file and output directory for batch generation.

8. Validate permissions for database metadata access, especially schema catalog reads such as information_schema or pg_catalog where applicable.

The skill organizes its data around database metadata, generation requests, SQL outputs, and operational audit artifacts.

Input parameters:

• content: primary request text
• mode: optional processing mode such as json, text, or markdown
• style: output style reference

Output structure is typically JSON-shaped with success flags, result payloads, metadata, and error fields. Batch workflows also rely on prompt files such as YAML, and generated files are commonly written into an output directory using a predictable naming convention.