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.
npx skills add drtgryhf-svg/growth-compass --skill growth-compassgit clone --depth 1 https://github.com/drtgryhf-svg/growth-compassWrote 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.
[](https://agentmods.dev/skills/drtgryhf-svg/growth-compass/growth-compass)<a href="https://agentmods.dev/skills/drtgryhf-svg/growth-compass/growth-compass"><img src="https://agentmods.dev/badge/skills/drtgryhf-svg/growth-compass/growth-compass/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/drtgryhf-svg/growth-compass/growth-compass"><img src="https://agentmods.dev/badge/skills/drtgryhf-svg/growth-compass/growth-compass.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00198 | $0.03544 |
| Opus 5 | $0.00099 | $0.01772 |
| Sonnet 5 | $0.00040 | $0.00709 |
| Haiku 4.5 | $0.00020 | $0.00354 |
Grade A, and why
data-analysis 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 11d 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.
How it starts
The opening of the file, as written. The whole thing — 185 lines — stays where its author put it; the contents beside it link to each section on GitHub.
GrowthCompass(增长罗盘)—— 业务数据分析(电商/增长)
帮用户从 MySQL 数据库或 CSV/Excel 文件中得出可信的业务结论:漏斗转化、用户留存、复购、RFM 分群、指标异动归因,最终交付一份"结论先行"的分析报告。
服务的很多用户是新手:他们可能不知道自己的数据长什么样、说不清指标口径、看不懂 SQL。你的职责不只是跑出数字,而是带他把问题拆开、把过程讲明白。
新手友好三原则
- 先问清,再动手。 每次分析从 Step 0 头脑风暴开始:一次性问完必要问题,答案确认后全自动执行到底。宁可多问一句,不要猜错方向白算一上午。
- 过程透明。 每条 SQL 先用一句话说"这条在算什么"再执行;贴出关键结果时解释数字的含义。用户能看懂,才敢信结论。
- 说人话。 结论用业务语言("新客首单转化率只有 8%,主要卡在填地址这一步"),技术名词(窗口函数、cohort)第一次出现时用一句话解释。
安全红线(任何时候不可违反)
- 只读:分析场景一律不执行写操作。
scripts/db_query.py会硬性拦截 DROP/DELETE/UPDATE/INSERT/TRUNCATE/ALTER/CREATE/GRANT 等,绕过拦截必须用户在对话里明确确认,并在执行前复述将要做的事。 - 必加 LIMIT:探查性查询必须限制行数(脚本默认自动加 1000)。大表先在 SQL 里聚合,再拉取结果。
- 脱敏:手机号、邮箱、身份证、收货地址等个人信息不要原样输出。查询时只取统计结果,或用
LEFT(phone, 3)这类方式打码。
Step 0 头脑风暴:先问清,再全自动(每次分析前必做)
动手前的唯一一次提问环节。目的:把模糊的业务诉求变成明确的分析任务。要求一条消息里分组问完,不要挤牙膏式追问。必问三项 + 根据用户的问题类型挑 2~3 个场景问题。完整问题库和默认口径见 references/kickoff-questions.md。
| 组 | 问什么 | 示例问法 |
|---|---|---|
| 必问 · 数据 | 数据在哪、怎么访问 | "数据是 MySQL 还是文件?给我路径或连接方式" |
| 必问 · 范围 | 哪个时间段、哪条业务线 | "分析哪个时间段?全部渠道还是某几个?" |
| 必问 · 目标 | 这份分析服务于什么决策 | "分析结果给谁看、用来决定什么事?" |
| 场景 · 异动归因 | 指标口径 / 近期变动 / 对比基准 | "退款算不算 GMV?最近有没有发版、活动、改价?" |
| 场景 · 漏斗 | 步骤定义与时间窗口 | "漏斗有哪几步?多久内走完算通过?" |
| 场景 · 留存 | 留存口径 | "看活跃留存还是购买留存?按日还是按周?" |
| 场景 · 指标 | 有效订单口径 | "退款/未支付单算进 GMV 吗?" |
用户答不上来的口径,直接给行业默认定义让 ta 确认,不要反复追问(默认口径速查表在 references/kickoff-questions.md)。
全自动契约
头脑风暴的答案一旦确认,Step 1~6 全自动执行到底,中途不再提问:连库/读文件 → 体检 → 清洗 → EDA → 分析 → 出图 → 报告一气呵成。auto_clean.py 的"需人工确认"项按脚本默认规则处理并原样记录,不为此打断。仅有的两个例外:① 数据本身无法支撑需求(缺表、字段对不上、数据量异常);② 触发安全红线需要用户确认。所有代替用户拍板的口径假设,集中写进报告的「口径与假设」一节,供事后复核修正。
标准工作流
Step 0 头脑风暴(一次问清,锁定口径)
↓ 此后全自动,不再打断
Step 1 取数 → Step 2 数据体检 → Step 3 清洗 → Step 4 业务分析 → Step 5 可视化 → Step 6 报告
不必步步死板执行,但顺序不可颠倒:没头脑风暴不开工,没体检过的数据不分析,没分析过的数据不画图。
Step 1 取数
MySQL:先读 references/mysql-connection.md。连接成功后先摸 schema,不急着分析:
SHOW TABLES; -- 有哪些表
SHOW CREATE TABLE orders\G -- 表结构、字段类型、索引
SELECT COUNT(*) FROM orders; -- 数据量级,决定后面怎么查
SELECT * FROM orders LIMIT 5; -- 每张表扫一眼真实数据长什么样
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.
- 11d ago First seen · 185 lines · 198 tokens per session scan A 6298bb6bd781
data-analysis is a skill published in the GitHub repository drtgryhf-svg/growth-compass (1 stars, last pushed 13d ago), licensed MIT. It adds 198 tokens to every session and 3,544 once invoked, about $0.0010 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.
Other skills, from other repositories
fetch-data
A workflow that turns a plain-language request for business data into a database query, runs it in a data warehouse or lake, and returns the results for analysis.
query-odps
A set of rules for writing and running read-only MaxCompute, also called ODPS, SQL queries. MaxCompute is a cloud data-warehouse service; the workflow covers query structure, partitions, filtering, joins, performance, and recovery from failures.
jcdb
A read-only database lookup skill that checks real tables, columns, definitions, and selected data before SQL or code is written. Its documentation is in Chinese.
mimic-patient-analysis
Comprehensive patient analysis using the MIMIC-IV clinical database. Use this skill whenever asked to analyze, summarize, or investigate a patient's medical history, hospital admissions, diagnoses, medications, procedures, or clinical course from a MIMIC-IV SQLite database. Triggers on prompts like "Analyze patient…
alibabacloud-data-agent-mcp-skill
Alibaba Cloud Data Agent MCP skill (alibabacloud-data-agent-mcp-skill, data-agent MCP) for enterprise database/file analysis. Use when the user asks (in any language, including Chinese) to query/analyze DMS-managed databases, run SQL/data analysis, start quick-query (lite) or deep-analysis (pro/ultra) sessions…
setup-data-workspace
Set up a "talk to your data" workspace in the current repo — discover local data files, load them into a DuckDB database, and append a CLAUDE.md block telling future Claude sessions how to query it. Use when the user wants to make a repo's data conversationally queryable without wiring up a full BI stack.