易经占卜 I Ching Divination

基于《周易》铜钱法和蓍草法起卦，生成本卦、互卦、变卦，提供诗性AI Oracle声音诠释用户提问。

This skill is largely coherent with its stated purpose (local I Ching divination) and contains only a small Python script plus data files, but there are multiple documentation mismatches you should resolve before installing or running it. What to check before installing/use: - Verify which hexagram data file is authoritative: SKILL.md references references/hexagrams.json (which is missing), README and references/hexagrams_metadata.json claim full 64, and SKILL.md claims only 10 examples. Open assets/hexagrams_full.json to confirm how many entries and whether the content is complete and accurate. - Confirm the script uses only local files and no network: review scripts/divine.py (it appears to use only the standard library and prints JSON). This script does not perform network requests or read external secrets. - Fix or be aware of doc/path typos: SKILL.md's file path should be updated to point to assets/hexagrams_full.json (or ensure references/hexagrams.json exists). Inaccurate docs (e.g., README saying "no file IO") reduce trustworthiness. - Run the script in a sandboxed environment first (non-production account) and inspect its output. If reproducibility matters, note that the script seeds randomness with hash(question) which may vary between Python runs due to Python's randomized hash seed — supply an explicit seed if you want deterministic results. If you need higher assurance, ask the maintainer (or request an updated package) to correct the documentation inconsistencies, include a clear mapping of where the 64 hexagrams are stored, and confirm there are no hidden network calls or extra install steps.

Type: OpenClaw Skill Name: yijing-divination Version: 1.0.0 The skill is classified as suspicious due to a critical shell injection vulnerability in `SKILL.md`. The instructions for the AI agent, `result=$(python3 scripts/divine.py --method coin --question "{用户问题}")`, directly interpolate user input (`{用户问题}`) into a bash command. If the OpenClaw agent does not properly sanitize or quote this input, a malicious user could inject arbitrary shell commands, leading to Remote Code Execution (RCE). While the Python script (`scripts/divine.py`) itself is benign and performs no external I/O or network calls, the method of execution described in `SKILL.md` presents a significant security risk. The `README.md`'s claim of 'No code execution risk' is therefore inaccurate in the context of the agent's execution model.