From ship
Handles plan divergence during build execution with three escalation levels: Fix and Continue, Fix with Limits, and Stop and Report.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ship:deviation-rulesThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
During build execution, follow these rules when the plan does not match reality.
During build execution, follow these rules when the plan does not match reality.
Trigger: A file path, function name, minor implementation detail, or missing dependency in the plan is wrong or outdated.
Action: Make the correct change (fix the path, install the dependency, adjust the detail). Continue executing the next task.
Examples:
src/auth/login.ts but the file is src/auth/auth.ts — fix it, move on.import bcrypt but bcrypt isn't installed — install it, move on.Trigger: The <verify> command for a task fails after implementation.
Action: Debug systematically, then fix and re-verify. Maximum 3 attempts. Track each attempt.
Debugging protocol (before each fix attempt):
If verify still fails after the third attempt, escalate to Rule 3.
If each fix reveals a new problem in a different place: This is not a single bug — it's an architectural mismatch. Skip directly to Rule 3.
Do not: Skip the verify step. Do not proceed to the next task with a broken current task. Do not guess — read the error first.
Trigger: An architectural conflict, persistent verification failure (3 attempts exhausted), or any issue that requires fundamentally rethinking the approach.
Action:
building in CONTEXT.md## CHECKPOINT REACHED with a clear explanation and recommendationSignals:
npx claudepluginhub dilhanz/ship --plugin shipGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.