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 agentmods add instructions/ifoodsci-ai/mflowy/agents-mdgit clone --depth 1 https://github.com/ifoodsci-ai/mflowyWrote 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/instructions/ifoodsci-ai/mflowy/agents-md)<a href="https://agentmods.dev/instructions/ifoodsci-ai/mflowy/agents-md"><img src="https://agentmods.dev/badge/instructions/ifoodsci-ai/mflowy/agents-md.svg" alt="Measured on agentmods" 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 | $0.01856 | $0.01856 |
| Opus 5 | $0.00928 | $0.00928 |
| Sonnet 5 | $0.00371 | $0.00371 |
| Haiku 4.5 | $0.00186 | $0.00186 |
Grade A, and why
mflowy AGENTS.md 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 3d 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 — 93 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
AI 编码 agent 在本仓库的唯一指引。代码是真相源——本文件只写代码里查不到的东西：约定、理由、陷阱、流程。存疑先探索再假设；动态文件树用 tree --gitignore 现查。
Project
MFlowy：YAML 配置驱动的模块化 ML 工作流引擎，全部能力以 MCP tools（pyfunc）暴露——MCP server（stdio）、JSON runner（K8s Job 容器）、直接 import 三入口。
Language: Python 3.12 | 代码注释与文档：中文
开发命令
uv sync --all-extras --all-groups # 首次/依赖变更后
make test # 全量测试
make lint && make fmt # ruff
uv run cmd list_modules # base 能力（无数据栈）
uv run --extra modeling cmd modeling '{"modeling_steps_yaml": "...", "name": "...", "desc": "..."}'
uv run mcpSrv # MCP server（stdio）
用户态分发（PyPI/uvx/wheel/Docker）见 README 与 docker/；MCP 客户端 stdio 配置见 .mcp.json.example。
uvx 陷阱（uvx 不读 pyproject 的 [tool.uv.sources]）：
- torch CPU 索引（
--index https://download.pytorch.org/whl/cpu+unsafe-best-match）为 [modeling] 必需——缺省时 torch 解析为 CUDA 全家桶（2–3GB） - extras 内联进
--from "mflowy[modeling]"——uvx 的--extra需新版 uv - 入口名
mcpSrv刻意避开 mcp SDK 同名 CLI（mcp.cli:app）——uvx 解析mcp命令可能命中 SDK 侧启动失败
分层
| 层 | 位置 | 职责 |
|---|---|---|
| MCP Tools | src/mflowy/mcp/tools.py |
全部工具定义，纯委派 |
| JobProvider | src/mflowy/mcp/job_provider/ |
执行策略（Local 直调 / Remote K8s Job） |
| Driver | src/mflowy/driver/ |
YAML 解析、DAG 调度、handler 派发、模块发现 |
| Compute | src/mflowy/compute/ |
业务逻辑 |
工具三分：建模类 4 种（modeling / explanation / predict / inverse_optimization）经 JobProvider 委派；分析类 3 种（data_profile / eda / infer_task_type_by_statistic）与 info/mlflow 组始终本地执行。
- 零手工注册：compute 模块入口函数加
@handler(StepType.X, ...)即完成注册（driver/discover.py启动扫描mflowy.compute.**；_前缀与_EXCLUDED词干除外） - 通道边界：
print（stdout）= 业务数据，per-task 捕获进NodeResult.output；logger（stderr）= 过程诊断。同一信息只走一条通道 - 模块内相对导入，跨模块绝对导入；KISS
tests/镜像src/mflowy/路径，integration/端到端
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.
- 3d ago First seen · 93 lines · 1,856 tokens per session scan A 3d207689a10c
mflowy AGENTS.md is an instructions file published in the GitHub repository ifoodsci-ai/mflowy (0 stars, last pushed 5d ago), licensed MIT. It adds 1,856 tokens to every session, about $0.0093 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 instructions, from other repositories
mlflow CLAUDE.md
Claude Code instructions for mlflow/mlflow, covering claude.md, knowledge cutoff note, code style principles, repository overview and quick start: development server.
mlflow code-review.instructions.md
Instructions for mlflow/mlflow, covering code review instructions, knowledge cutoff and do not comment on.
mlflow github-actions.instructions.md
Instructions for mlflow/mlflow: For workflow style conventions, see .claude/rules/github-actions.md.
mlflow python.instructions.md
Instructions for mlflow/mlflow: For style conventions and code examples, see .claude/rules/python.md.
mcp-datahub CLAUDE.md
Claude Code instructions for txn2/mcp-datahub, covering claude.md, project overview, project positioning, key design principles and architecture.
mlflow AGENTS.md
AGENTS.md instructions for mlflow/mlflow, a project described as: The open source AI engineering platform for agents, LLMs, and ML models. MLflow enables teams of all sizes to debug, evaluate, monitor, and optimize production-quality AI applications while controlling costs and managing access to models and data.