Taxonomy Normalizer

统一不同团队、不同表里的分类体系，保留别名映射与废弃词。；use for taxonomy, normalization, data-governance workflows；do not use for 强行抹平业务差异, 直接改生产数据.

This skill appears coherent and low-risk for its stated purpose, but take normal precautions before running it: (1) Inspect scripts yourself (scripts/run.py is included) and run in a safe/test environment first. (2) Use --dry-run or redirect output to a temporary file before applying any suggested changes. (3) Avoid pointing the tool at directories containing unredacted secrets or sensitive PII—the script will read and summarize many file types and may surface sensitive values in reports. (4) Confirm you trust the skill source before giving it large or production directories to scan. If you need networked or automated writes to external systems, require additional review and explicit safeguards—this skill is intended to produce reviewable suggestions, not to change production systems.

Type: OpenClaw Skill Name: taxonomy-normalizer Version: 1.0.0 The skill bundle is a well-documented tool for taxonomy normalization and data governance. The core logic in `scripts/run.py` is robust and includes several auditing modes (e.g., `pattern_audit`, `skill_audit`) that appear to be part of a reusable framework; notably, the pattern scanner includes logic to redact detected secrets rather than exfiltrate them. The instructions in `SKILL.md` are safety-conscious, explicitly advising the agent against modifying production data or executing high-risk commands. No evidence of malicious intent, data exfiltration, or harmful prompt injection was found.