Scan Report
0 /100
paper-to-pipeline
根据机器学习/深度学习论文的实验规划文档自动生成完整的 Python 实验 pipeline。支持数据预处理、模型构建、训练循环、评估指标、结果可视化。
A clean ML pipeline code generator with no security issues; reads experiment plans and writes standard Python ML code to a user-specified directory using only standard libraries.
Safe to install
No action needed. The skill is safe to use as described.
| Resource | Declared | Inferred | Status | Evidence |
|---|---|---|---|---|
| Filesystem | READ | READ | ✓ Aligned | scripts/generate_pipeline.py reads plan files via open() |
| Filesystem | WRITE | WRITE | ✓ Aligned | SKILL.md declares output to generated_experiment/ directory; code writes to user… |
| Network | NONE | NONE | — | No network calls in any script |
| Shell | NONE | NONE | — | No subprocess/os.system/os.popen calls found |
| Environment | NONE | NONE | — | No os.environ access in generate_pipeline.py |
| Skill Invoke | NONE | NONE | — | No skill invocation patterns detected |
| Clipboard | NONE | NONE | — | No clipboard access |
| Browser | NONE | NONE | — | No browser/web automation |
| Database | NONE | NONE | — | No DB connections |
File Tree
5 files · 52.0 KB · 1951 lines Python 2f · 1224L
Markdown 3f · 727L
├─
▾
assets
│ └─
▾
templates
│ └─
image_classification.py
Python
├─
▾
references
│ ├─
data-format.md
Markdown
│ └─
model-templates.md
Markdown
├─
▾
scripts
│ └─
generate_pipeline.py
Python
└─
SKILL.md
Markdown
Security Positives
✓ No shell execution (subprocess, os.system, curl|wget, or pip install commands)
✓ No network requests — all imports are local standard library (argparse, os, re, yaml, pathlib, datetime)
✓ No credential harvesting — no access to ~/.ssh, ~/.aws, .env, or environment variable enumeration
✓ No obfuscation — no base64, no eval(atob()), no encoded payloads
✓ No persistence mechanisms — no cron jobs, startup scripts, or backdoor installation
✓ No prompt injection — generated code contains only standard ML boilerplate
✓ No sensitive file paths accessed
✓ Doc-to-code match is excellent — SKILL.md accurately describes all behavior
✓ All file writes are to user-specified output directories, declared in SKILL.md
✓ Dependencies are standard ML libraries (torch, torchvision, transformers) only in generated requirements.txt, not in the skill itself