vibe-bump

vibe-bump is a skill for Claude Code from chentao326/vibe-code. It costs 71 tokens per session (1,135 once invoked), scanned A, original, MIT.

A workflow for revising the scoring model, or rubric, used to evaluate coding tasks. It requires enough completed examples, rescoring those examples, checking whether their order stays consistent, and getting an independent review.

In plain words
What is it for?
Updating rubric dimensions or weights after calibration, comparing old and new task rankings, reviewing the proposed formula across past samples, and applying an approved model change.
Why use it?
It prevents changing evaluation rules based on too little evidence or on a single opinion.

Skill for Claude Code

Written for Claude Code: allowed-tools in frontmatter. Also seen: mentions subagents.

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/chentao326/vibe-code/vibe-bump
Any agent
npx skills add chentao326/vibe-code --skill vibe-bump
Clone the repo
git clone --depth 1 https://github.com/chentao326/vibe-code

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for vibe-bump

README.md
[![agentmods](https://agentmods.dev/badge/skills/chentao326/vibe-code/vibe-bump.svg)](https://agentmods.dev/skills/chentao326/vibe-code/vibe-bump)
Your own site
<a href="https://agentmods.dev/skills/chentao326/vibe-code/vibe-bump"><img src="https://agentmods.dev/badge/skills/chentao326/vibe-code/vibe-bump.svg" alt="Measured on agentmods" height="20"></a>
Per session 71 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,135 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00071 $0.01135
Opus 5 $0.00036 $0.00567
Sonnet 5 $0.00014 $0.00227
Haiku 4.5 $0.00007 $0.00113

Measured 5d ago against content hash ea36c35cbe81, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

vibe-bump scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/vibe-bump/SKILL.md · 111 lines

How it starts

The opening of the file, as written. The whole thing — 111 lines — stays where its author put it; the contents beside it link to each section on GitHub.

vibe-bump — 升级评估模型

严格遵守 bump-validation-protocol.md


Overview

Phase 0: 前置门槛检查
Phase 1: 写出新公式完整方程
Phase 2: 校准池全量重打分(强制 blind sub-agent)
Phase 3: 计算排序一致性
Phase 4: 跨模型独立审核
Phase 5: 落地 + cleanup

Constants

  • THRESHOLD = 0.8 — 排序一致性阈值(4/5),刚性不可改
  • CROSS_MODEL_AUDIT = true — 默认开启跨模型审核
  • MIN_SAMPLES_FOR_BUMP = 5 — 首次 bump 最低门槛

Workflow

Phase 0: 前置门槛检查

检查项 失败处理
calibration_samples ≥ 5? 不满足 → 说明原因 + 拒绝(除非强反例如 ≥3x 偏差,标 judgment-driven)
有 ≥1 个新样本自上次 bump? 不满足 → 拒绝
in_progress_assessment == null? 不满足 → 拒绝
有清晰偏差方向? 询问用户为什么现在 bump

通过 → 进入 Phase 1。

Phase 1: 写出新公式完整方程

用户描述要改什么 → AI 帮助写出完整的方程。

示例:

v0: composite = (CS×1.0 + CX×1.0 + AM×1.0 + TE×1.0 + AQ×1.0) / 5 × 2.0
v1: composite = (CS×1.5 + CX×1.0 + AM×1.0 + TE×1.5 + AQ×1.0) / 6 × 2.0

不允许只说"CS 权重提高"——必须完整方程。

Phase 2: 校准池全量重打分

校准池 = 所有 predictions/*.md 有完整复盘段的文件。

强制走 blind sub-agent(不接受 self-scored fallback):

  1. 对每个校准样本,Task tool spawn sub-agent
  2. sub-agent 只读 task 原文件 + 新公式 rubric.md
  3. sub-agent 输出新的 N 维分
  4. 用新公式重新计算 composite

sub-agent 禁读:所有 prediction 文件、retro 段、state 文件——只读 task 原文和新 rubric。

Phase 3: 计算排序一致性

对新 composite 排序 vs 实际"效率"排序:

效率定义:1 / (耗时_min × 迭代次数 × (bug数 + 1))

  • 计算 Spearman rank correlation
  • 检查 pairwise 顺序是否颠倒
  • 满足 ≥4/5 样本排序一致 → 通过

Phase 4: 跨模型独立审核

打包以下发给外部 LLM:

  • 旧公式 + 新公式
  • 校准池数据(每样本:维度分、新旧 composite、实际耗时/轮次/bug)
  • Step 3 排序对照表

外部 LLM 判定:PASS / REJECT + ≥100 字理由。

本地 + 外部都通过 → Step 5。

Phase 5: 落地 + cleanup

  1. 更新 rubric.md 顶部版本 + 公式
  2. 写入"升级 Memo"段(触发原因/证据/新公式/局限)
  3. 删除被吸收为维度或被推翻的观察(observation-lifecycle.md)
  4. 未解决观察 → 迁移到"待验证假设"
  5. 所有校准样本 prediction 文件底部追加 **Re-scored under v1 on YYYY-MM-DD**: composite=X.XX → Y.YY
  6. 更新 state: rubric_version += 1, last_bump_at = now, calibration_samples_at_last_bump = current

Bump 被拒处理

失败位置 处理
Step 3 排序不一致 候选公式回退。不允许放宽 THRESHOLD
Step 4 外部 REJECT 记录外部理由到 rubric.md
Step 4 双方冲突 视为 REJECT

Read the full file on GitHub · 111 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 5d ago First seen · 111 lines · 71 tokens per session scan A ea36c35cbe81

Subscribe to this mod's changes

vibe-bump is a skill published in the GitHub repository chentao326/vibe-code (2 stars, last pushed 3mo ago), licensed MIT. It adds 71 tokens to every session and 1,135 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

local-ai-agents

Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…

microsoft/ai-agents-for-beginners · 200 tokens

next-cache-components-adoption

Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…

vercel/next.js · 95 tokens

next-partial-prefetching-adoption

Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the partialPrefetching flag, opt routes in with export const prefetch = 'partial', audit Link prefetch={true} behavior, preserve existing prefetched UI with…

vercel/next.js · 103 tokens

chronicle

Analyze Copilot session history for standup reports, usage tips, session search, and session reindexing. Use when the user asks for a standup, daily summary, usage tips, workflow recommendations, wants to search or find past sessions by keyword/file/PR, wants to reindex their session store, or asks about deleting…

microsoft/vscode · 72 tokens

imagegen

Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…

openai/codex · 113 tokens