中文 EN

扫描报告

扫描新文件

✓ 可信 — 风险评分 0/100
上次扫描：1 天前 重新扫描
0 /100
dr-context-pipeline
Deterministic memory/context pipeline for agents: route a user message, retrieve relevant memory snippets, compress into a cited Context Pack, lint, and fall back safely.
A clean, legitimate memory/context pipeline skill with no malicious behavior, no undeclared permissions, and documentation that accurately reflects implementation.
技能名称dr-context-pipeline
分析耗时26.4s
引擎pi
✓
可以安装
No action needed. The skill is safe to deploy.
资源类型声明权限推断权限状态证据
文件系统 WRITE WRITE ✓ 一致 install_pipeline.py:34 — shutil.copytree for targeted pipeline installation
网络访问 NONE NONE — No network calls in any script
命令执行 NONE NONE — No subprocess/os.system calls in any script
环境变量 NONE NONE — No os.environ access in any script
技能调用 NONE NONE — No dynamic skill invocation
剪贴板 NONE NONE — No clipboard access
浏览器 NONE NONE — No browser automation
数据库 NONE NONE — No database access
1 项发现
🔗
中危 外部 URL 外部 URL
https://json-schema.org/draft/2020-12/schema
assets/context_pipeline/schemas/context_pack.schema.json:2

目录结构

20 文件 · 49.5 KB · 1455 行
JSON 6f · 518L YAML 2f · 344L Python 3f · 308L Markdown 7f · 243L Text 2f · 42L
├─ ▾ 📁 assets
│ └─ ▾ 📁 context_pipeline
│ ├─ ▾ 📁 schemas
│ │ ├─ 📋 context_pack.schema.json JSON 119L · 3.3 KB
│ │ └─ 📋 retrieval_bundle.schema.json JSON 69L · 2.0 KB
│ ├─ ▾ 📁 tests
│ │ └─ 📋 golden.json JSON 56L · 1.4 KB
│ ├─ 📝 always_on.md Markdown 25L · 1.5 KB
│ ├─ 📄 compressor_prompt.txt Text 21L · 1.1 KB
│ ├─ 📝 README.md Markdown 20L · 919 B
│ └─ 📋 router.yml YAML 172L · 4.0 KB
├─ ▾ 📁 references
│ ├─ ▾ 📁 schemas
│ │ ├─ 📋 context_pack.schema.json JSON 119L · 3.3 KB
│ │ └─ 📋 retrieval_bundle.schema.json JSON 69L · 2.0 KB
│ ├─ ▾ 📁 tests
│ │ └─ 📋 golden.json JSON 86L · 2.2 KB
│ ├─ 📝 APPLY.md Markdown 34L · 2.1 KB
│ ├─ 📄 compressor_prompt.txt Text 21L · 1.1 KB
│ ├─ 📝 configure_prompt.md Markdown 38L · 2.1 KB
│ ├─ 📝 deterministic_ids.md Markdown 22L · 775 B
│ ├─ 📋 router.yml YAML 172L · 4.0 KB
│ └─ 📝 RUNTIME_CHECKLIST.md Markdown 29L · 1.6 KB
├─ ▾ 📁 scripts
│ ├─ 🐍 install_pipeline.py Python 69L · 2.0 KB
│ ├─ 🐍 memory_watchdog.py Python 103L · 3.4 KB
│ └─ 🐍 validate_pipeline.py Python 136L · 4.7 KB
└─ 📝 SKILL.md Markdown 75L · 6.0 KB

安全亮点

✓ SKILL.md accurately describes all script behavior with no hidden functionality
✓ No network requests, credential access, or data exfiltration in any script
✓ install_pipeline.py only copies files to a specific target directory using shutil.copytree
✓ validate_pipeline.py only hashes and compares files, reads JSON schemas — purely local operations
✓ memory_watchdog.py only checks file mtimes and sizes in the memory directory — no writes or network
✓ No subprocess, os.system, or shell execution anywhere
✓ No base64, eval, or obfuscation patterns
✓ No sensitive paths accessed (~/.ssh, ~/.aws, .env)
✓ Dependencies: only Python stdlib (hashlib, json, shutil, pathlib, dataclasses, datetime, argparse) — no third-party packages required
✓ Version not pinned for stdlib — irrelevant since stdlib has no external attack surface
✓ Scripts are deterministic and read-only after installation (watchdog only reads mtimes/sizes)