wjs-evaling-voicedrop-prompts

Solid

Use when 王建硕 wants to evaluate whether a change to VoiceDrop's 挖矿 system prompt is actually better than the live version — runs the local eval harness (golden fixtures × champion-vs-candidate, same input), dispatches blind pairwise judge subagents, aggregates a win-rate verdict, and on approval promotes the candidate into agent/src/prompts/mine.js. Triggers — "评估 prompt"、"挖矿 prompt 改好了吗"、"eval prompt"、"比一比两版 prompt"、"/wjs-evaling-voicedrop-prompts".

AI & Automation 129 stars 21 forks Updated 3 weeks ago MIT

Install

View on GitHub

Quality Score: 83/100

Stars 20%
70
Recency 20%
90
Frontmatter 20%
70
Documentation 15%
69
Issue Health 10%
50
License 10%
100
Description 5%
100

Skill Content

# VoiceDrop 挖矿 prompt 评估 harness = 本 skill(协议)+ jianshuo.dev `agent/eval/`(脚本/数据)。运行时 = 本地 Claude Code。 被测对象 = `agent/src/prompts/mine.js` 的 `MINE_SYSTEM`(git 即版本库)。 ## 何时用 用户改了挖矿 prompt(`MINE_SYSTEM`),想用数据判断改好了还是改坏了,而不是凭感觉看一两次。 ## 流程(按序) 1. **拿候选 prompt**:把候选版 `MINE_SYSTEM` 文本写到一个临时文件(如 `/tmp/cand-prompt.txt`);冠军 = 当前 `mine.js` 的 `MINE_SYSTEM`(脚本自动读)。 2. **跑产出**:`cd ~/code/jianshuo.dev/agent && CLAUDE_API_KEY=$CLAUDE_API_KEY node eval/run-eval.mjs /tmp/cand-prompt.txt <runId>`。产出落 `eval/runs/<runId>/`。先看终端有没有「确定性回退」警告——有就先停,多半是候选 prompt 破坏了 JSON 输出。 3. **成对盲评**:对每条 fixture,dispatch 一个 subagent,喂 `references/judge-rubric.md` + 该 fixture 的 transcript + 两份产出。**A/B 顺序随机**(一半 fixture 把 candidate 放 A、一半放 B,记录映射,收到结果后还原成 champion/candidate)。裁判模型用与生成(opus)不同家族的模型。收每条的 `{winner, dims, reason}`。 4. **聚合**:把还原后的 `verdicts`(winner ∈ candidate/champion/tie)+ 候选 proxyFails 喂 `aggregate()`,渲染 `renderReport()` → 写 `eval/runs/<runId>/report.md`。 5. **人工终审**:把胜负最接近、分歧最大的 1–2 条产出并排摆给用户。**机器只筛掉明显更差的,文风最后一票是用户。** 6. **晋级**:仅当 `decision==="promote"`(胜率 ≥70% 且无回退)**且用户点「认可」**——把候选写回 `agent/src/prompts/mine.js` 的 `MINE_SYSTEM`,commit(message 附 runId 与胜率),并跑 `npm test` 确认没破坏。否则保留报告、不动生产版。 ## 边界 - 只评挖矿 prompt(`MINE_SYSTEM`/`MINE_SYSTEM_FORCE`)。审核/语音编辑 prompt 是不同 eval 模式,不在本 skill。 - 不测成本/缓存/延迟(本地缓存行为≠生产);不做无人值守。 - 真实金标集要 ≥10 条才可信(见 `agent/eval/fixtures/README.md` 的补充流程);种子 2 条只够自测流程。

Details

Author
jianshuo
Repository
jianshuo/claude-skills
Created
4 months ago
Last Updated
3 weeks ago
Language
Python
License
MIT

Integrates with

Bundled in these plugins

Similar Skills

Semantically similar based on skill content — not just same category

AI & Automation Solid

ai-evaluation-engineering

AI 评测工程专家 Owner — 当任务涉及模型/Prompt 评测、模型选择、黄金集、评分量表、LLM-as-judge、Judge 校准、重复采样、方差、质量-成本-延迟权衡、提示词回归或模型升级回归时使用;要求把概率性结果转化为可复现、可比较且防污染的评测证据。

263 Updated 2 days ago
devcodex-labs
AI & Automation Listed

devlab-eval-driven-agent

eval 自评测体系驱动的 AI Agent 生产体系方法论。以评测集为核心资产,建立"评测集组织 + Mock 隔离 + 标准化比对 + 自动评测脚本 + 回归门禁"闭环;L0-L4 分层评测(确定性单测/轨迹评测/输出评测/生产回归/安全成本护栏),L1 golden session 轨迹断言,L2/L3 数据后端接 Langfuse(dataset/scores/LLM-as-judge),配套 evalctl CLI(0.3.x)。Triggers on "评测集", "eval", "回归测试", "准确率怎么保证", "测试数据集", "AI 质量护栏", "eval-driven", "LLM-as-judge", "langfuse 评测", "scores", "质量评分", "轨迹评测", "golden session", "agent 测试", "工具调用序列".

22 Updated 1 weeks ago
seed-forge
AI & Automation Listed

ai_gym

Train AI agent prompts by MEASUREMENT instead of vibes — stand up an eval harness (real prompt + real LLM + STUBBED tools = zero side effects), author scenarios with pass/fail rubrics, judge transcripts with subagents, aggregate a scorecard by dimension, run a TOURNAMENT of prompt variants, kill LLM noise with N-runs, and ship only the change measurement proved. Use whenever the user wants to improve/refine/test the system prompt of an AI agent, chatbot, or LLM pipeline; compare two prompt versions; find why an agent hallucinates, leaks data, or escalates wrong; build a regression eval for a prompt; or run "agent battles / a prompt tournament". Triggers on requests like "make my agent's prompt better", "this bot keeps making things up", "which prompt version is better", "write a test for my bot's behavior", "I need an eval for my agents" — even without the word "eval". Target stack: TS/Node + vitest; the LLM provider is pluggable (your project supplies the client). Works with any subagent-capable coding agent

0 Updated 1 months ago
luisroquette