spec-discipline

What problem does it solve?

This Skill eliminates the costly waste of parallel workers executing correct code for incorrect bug diagnoses, which leads to multiple shipped fixes that pass all tests but fail to resolve the user's actual reported symptom, a pattern that has caused production sagas lasting 5+ releases.

Core Features & Use Cases

• Falsifiable Hypothesis Specs: Enforces bug specifications that list ranked, testable hypotheses instead of unproven conclusions, allowing workers to challenge the director's framing before writing code.
• Mandatory Phase-0 Data Audits: Requires a physical data location check for all state, persistence, and sync bugs before any fix work is delegated, preventing fixes based on incorrect assumptions about where data lives.
• Two-Fix Stop Rule & Task Force Protocol: Freezes incremental fixes when a bug survives two shipped releases, dispatching a cross-functional team to audit data, deployment, end-to-end traces, and live reproductions to find the real root cause.
• Premise-Challenger Cross-Validation: Assigns one reviewer per cross-validation cycle to attack unverified assumptions in the bug spec, preventing "CLEAN theater" where PRs pass review but are built on false premises.
• Use Case: Use this Skill when a user-reported bug has had multiple CLEAN PRs that do not fix the symptom, when writing bug specs for worker dispatch, or when reviews keep passing but the user still sees the issue.

Quick Start

Use the spec-discipline skill to restructure your current bug spec for the unresolved recurring user symptom to include falsifiable hypotheses, a mandatory phase-0 data audit step, and a premise-challenger review requirement.