菜单
Detects and prevents shortcut rationalization — when the agent is about to skip a step, simplify a requirement, or optimize away a constraint. Triggers proactively when rationalization patterns are noticed.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Dispatches research agents to find real projects, papers, and documented failures before proposing any feature, architecture, or technology choice. Triggers on creative/design tasks, not on implementation of already-designed specs.
Validates completed work against defined acceptance criteria. Use after completing a task that has specific success criteria defined in issues, specs, or task descriptions.
Generates edge case, failure mode, and spec-driven test cases. Covers boundary values, nil inputs, concurrency, resource exhaustion, malformed data, and requirement-linked traceability tests.
Provides a persistent cross-session task queue for work that should be done but isn't blocking current tasks. Enables background and future-session work tracking.
Safely integrates commits from parallel agent branches using sequential cherry-pick. Use after parallel work completes in isolated branches or worktrees.
Audits tests for concurrency safety — race conditions, shared mock state, cleanup ordering. Use when writing tests that involve goroutines, async operations, or shared mutable state.
| name | vibe-anti-rationalization-check |
| description | Detects and prevents shortcut rationalization — when the agent is about to skip a step, simplify a requirement, or optimize away a constraint. Triggers proactively when rationalization patterns are noticed. |
| user-invocable | true |
LLMs systematically rationalize shortcuts. This skill catches it happening.
Watch for these thoughts — they're red flags:
| Rationalization | Reality |
|---|---|
| "This is too simple to need tests" | Simple code in complex systems causes the hardest bugs |
| "The user probably doesn't care about this edge case" | Production users will find every edge case |
| "We don't need error handling here" | Internal code fails too, especially during refactors |
| "I'll add that later" | "Later" never comes. Do it now or create a tracked task |
| "This is just boilerplate" | Boilerplate often contains critical correctness constraints |
| "The happy path is enough for now" | The sad path is where real users spend most of their time |
| "It works, so it must be correct" | "Works" and "correct" are different things |
| "This test is redundant" | Redundant tests catch redundant bugs |
| "Let me simplify this requirement" | Simplifying without asking = silently dropping scope |
| "This doesn't need documentation" | If you thought about it for >2 minutes, document why |
When you detect a rationalization:
This skill doesn't produce output — it's a mental discipline check. When triggered, it modifies behavior, not output.
If invoked explicitly by the user, report:
