rdd-defect-workflow

Guides receipt-driven defect fixes with issue gates, evidence journeys, and bounded review validation.

6.1k|707|Updated Feb 27, 2026
One-click install
npx skills add https://github.com/Gentleman-Programming/gentle-ai --skill rdd-defect-workflow
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: rdd-defect-workflow
Source: https://github.com/Gentleman-Programming/gentle-ai/tree/main/internal/assets/skills/rdd-defect-workflow
Command: npx skills add https://github.com/Gentleman-Programming/gentle-ai --skill rdd-defect-workflow

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Fixing defects in a Receipt-Driven Development (RDD) repository requires strict evidence discipline: approved issues, reproduction on main, causal invariant isolation, line budgets, and independent validation. This Skill enforces that workflow so contributors never fabricate approvals, merge conflicting authority lines, or ship unproven fixes.

Core Features & Use Cases

  • RDD Mode and Gate Enforcement: Checks the user-owned RDD kill switch first, requires an approved issue and clean main reproduction, and audits existing PRs for supersession or conflict before any work begins.
  • Evidence-Bound Correction: Groups work by causal invariant, inventories every operator flow with negative controls, demands truthful black-box bench journeys or real runtime E2E proof, and enforces a 400-line change budget with chain or exception handling.
  • Independent Validation: Requires read-only candidate validation before publication and returns a structured output contract covering mode, invariants, flows, evidence, budget, tests, rollback, and unresolved decisions.
  • Use Case: A maintainer asks the agent to fix a CLI defect in an RDD-enabled repository; the Skill walks through mode check, issue gate, worktree isolation, failing tests, candidate freeze, and read-only validation before reporting the verdict.

Quick Start

Ask the agent to fix an approved defect issue in this repository following the RDD defect workflow with full evidence and validation.

Frequently Asked Questions about rdd-defect-workflow

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fix a defect in a receipt-driven development repository?

Start by checking the RDD kill switch and confirming an approved issue with clean reproduction on main. Then isolate a worktree, map impact with CodeGraph, add failing tests, apply the smallest correction, and require independent read-only validation before publication.

What happens when RDD mode is disabled for a defect fix?

When the RDD kill switch is disabled, the workflow reports disabled/unmanaged and follows ordinary repository policy. No receipt reviews are started and no approval is fabricated.

What is the change size limit for an RDD defect fix?

The hard limit is 400 additions plus deletions, forecast before any edits. If the forecast exceeds it, work must stop and proceed as a chain of smaller changes or obtain an explicit maintainer-approved exception.

Does synthetic test coverage prove a fix in another runtime?

No. Each CLI or lifecycle flow needs one truthful black-box bench journey, or actual runtime E2E proof when the core bench cannot represent it. Synthetic proxy coverage never proves another runtime.

Can one PR fix multiple independent defect causes?

No. Work is grouped by causal authority invariant, with one issue and one PR or explicit chain per independent invariant and rollback boundary. Independent causes must be split, and superseded or conflicting lines are never merged.