arch-diagram
v1.0.1生成代码仓架构可视化图，输出为独立静态 HTML 网页。 仅支持在代码仓根目录下运行，自动扫描当前目录。 Claude 扫描代码文件、理解架构、生成层次思维导图和各模块子流程图， 最终输出可在浏览器直接打开的 HTML 文件。
⭐ 0· 52·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
Capability signals
These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.
OpenClaw
Benign
high confidencePurpose & Capability
Name/description match the included scripts and prompts: scan_repo.py enumerates files and stats, parse_mindmap.py extracts nodes, build_html.py renders an HTML page from model outputs. All requested/installed pieces are proportional to generating an architecture diagram from a code repository.
Instruction Scope
SKILL.md explicitly instructs reading the entire repository (files' contents), batching them, and sending those contents to the LLM using the provided prompt templates. That is coherent with the stated purpose, but it means full repo contents (or large portions) will be exposed to the model invocation. It also writes caches (/cache/), temporary files under /tmp/, and an output HTML — all expected. The prompts constrain output formats (plantuml, JSON, mermaid) and require reading references/prompts.md before processing (present in package).
Install Mechanism
No install spec; the skill is instruction-plus-scripts only. That minimizes install-surface risk because nothing is downloaded from external URLs or installed system-wide.
Credentials
The skill declares no required environment variables or credentials — consistent with local-only scripts. However, runtime behavior relies on the agent/model (Claude) to process file contents; that requires whatever model access the platform provides (not declared in skill metadata). Users should be aware that repository contents will be sent to the LLM provider when the agent runs these prompts.
Persistence & Privilege
always is false and scripts only write local cache, /tmp files, and an output HTML. The skill does not attempt to modify other skills or system configuration and does not require permanent presence.
Assessment
This skill will scan the repository in the current directory, read file contents, and send those contents in batches to the LLM (per the included prompt templates) to produce summaries, a PlantUML mindmap, and Mermaid subgraphs, then write results to cache/, /tmp/, and an output HTML. That behavior is expected for generating architecture diagrams, but:
- If your repo contains secrets, credentials, or proprietary code you do not want transmitted to an external model provider, do NOT run this skill against that repo (or remove/rotate secrets first).
- The generated HTML includes the model outputs embedded into a template; treat the file as untrusted content (open it in a secure environment), because model-generated content could include unexpected payloads.
- The package does not download external code or require additional system installs. Review references/prompts.md if you want to inspect exactly what is sent to the model.
If you are comfortable with the model receiving your repo contents (or you run it on sanitized/public code), the skill appears coherent and appropriate for its purpose.Like a lobster shell, security has layers — review code before you run it.
latestvk97adrva8g17tcsn3yxvd6f0t984tzkt
License
MIT-0
Free to use, modify, and redistribute. No attribution required.