Multi-agent DAG execution framework for decomposing large features into independently verifiable work units with quality gates.

Structures complex features as directed acyclic graphs with explicit dependencies, complexity tiers (isolated edits to schema/security changes), and per-unit acceptance criteria
Enforces a seven-stage pipeline: RFC intake, DAG decomposition, unit assignment, implementation, validation, merge queue processing, and final system verification
Implements merge queue rules preventing dependency failures, enforcing rebasing on the integration branch, and re-running tests after each merge
Includes recovery mechanisms for stalled units: eviction, findings snapshot, scope regeneration, and retry with updated constraints
Produces RFC execution logs, unit scorecards, dependency graphs, and integration risk summaries for visibility across the orchestration workflow

Ralphinho RFC Pipeline

Inspired by humanplane style RFC decomposition patterns and multi-unit orchestration workflows.

Use this skill when a feature is too large for a single agent pass and must be split into independently verifiable work units.

Pipeline Stages

Each work unit should include:

If a unit stalls: