recce

Progressive data review specialist for dbt model changes. Dispatched by /recce-review skill after dbt runs. Calls impact_analysis for data evidence, reads model SQL for root cause diagnosis, and validates findings against stakeholder intent to produce an actionable summary with risk level. <example> Context: Developer ran `dbt run` and the /recce-review skill dispatched this agent user: "Review the data changes from my recent dbt run" assistant: "I'll dispatch the recce-reviewer agent to run impact analysis and data review." <commentary> Post-dbt-run data review is the primary trigger for this agent. </commentary> </example> <example> Context: Developer explicitly wants to check data impact before committing user: "Check if my model changes have any data impact" assistant: "I'll use the recce-reviewer agent to run impact analysis." <commentary> Manual data review request before commit — another common trigger. </commentary> </example> <example> Context: Multiple models were changed and need validation user: "I changed stg_orders and fct_revenue, please review the data" assistant: "I'll dispatch the recce-reviewer agent focused on those models." <commentary> Named model list review — agent handles via selector construction. </commentary> </example>

Review dbt model data changes using Recce. Triggers when: user asks to review data changes, check data impact, run recce review, or validate model changes before committing.

4 hooks across 3 events