sqlfluff-project-onboarding

name	sqlfluff-project-onboarding
description	Use when onboarding to a SQL or dbt-style project and determining the right SQLFluff dialect, templater, config, and baseline lint commands.
compatibility	Requires sqlfluff on PATH.

SQLFluff Project Onboarding

Purpose

Establish how SQLFluff should run in an unfamiliar SQL or dbt-style repository before linting broadly or applying fixes. Use this to discover the local SQLFluff version, dialect, templater, rule surface, configuration files, and safe baseline commands.

Workflow

1. Confirm SQLFluff availability

   • Run sqlfluff version.
   • If SQLFluff is missing, stop and report that the project needs SQLFluff installed before this workflow can continue.

2. Inspect the available SQLFluff surface

   • Run sqlfluff dialects --nocolor -v.
   • Run sqlfluff rules --nocolor -v.
   • Note warehouse-relevant dialects such as bigquery, snowflake, databricks, redshift, postgres, duckdb, trino, and sparksql when available.

3. Discover project configuration

   • Look for SQLFluff config files such as .sqlfluff, pyproject.toml, setup.cfg, or tox.ini.
   • Look for dbt indicators such as dbt_project.yml, models/, macros/, packages.yml, and profiles.yml references.
   • Identify configured dialect, templater, rules, exclude_rules, ignore, and .sqlfluffignore behavior when present.

4. Check templater assumptions

   • Do not assume the dbt templater is installed. SQLFluff installations may only expose templaters such as raw, jinja, python, and placeholder.
   • If the project expects dbt templating but SQLFluff rejects --templater dbt, report the missing templater as an environment/setup issue.

5. Run a narrow baseline lint

   • Prefer one representative file or a small directory first.
   • Use --nofail, --format json, and --disable-progress-bar for agent-readable exploratory output.

6. Report the operating model

   • Summarize the detected dialect, templater, config files, ignore policy, and recommended first commands.
   • Call out uncertainty explicitly instead of guessing.

Command Patterns

sqlfluff version
sqlfluff dialects --nocolor -v
sqlfluff rules --nocolor -v
sqlfluff lint models/staging/orders.sql --nofail --format json --disable-progress-bar

Output

Return a concise onboarding summary with:

• SQLFluff version and availability.
• Detected or recommended dialect.
• Detected or recommended templater.
• Configuration files and ignore files found.
• A safe first lint command for the project.
• Any blockers, especially missing templater support.

Success Criteria

• The agent can run SQLFluff with an explicit dialect/config strategy.
• The agent avoids broad linting or fixing until the project setup is understood.
• The user can see exactly which assumptions are confirmed and which remain open.

Progressive Disclosure

• Additional CLI details: references/cli-cheatsheet.md