A股/炒股新手概念与指标的结构化讲解与笔记沉淀。用在：解释PE/PB/ROE/现金流/财报口径、板块与题材、政策术语；把零散问题整理成可复习的知识体系（中文优先）。

This skill appears coherent and low-risk: it generates explanations and appends them to notes/stocks/glossary.md locally and does not ask for credentials or install software. Before enabling: (1) confirm you are comfortable with the agent writing to notes/stocks/glossary.md (move or rename the path if you prefer another location); (2) review any appended notes for accuracy—the skill explicitly avoids giving buy/sell advice but may still be incorrect; (3) in multi-user or shared environments, consider file permissions or a private folder to avoid exposing appended notes; (4) if you pass links as context, the skill’s instructions do not say it will fetch external content automatically—assume it will not unless you see additional code or network instructions.

Type: OpenClaw Skill Name: a-share-glossary-tutor Version: 1.0.0 The skill instructs the agent to write user-derived content to a local markdown file (`notes/stocks/glossary.md`) for persistence, as detailed in `SKILL.md`. While the stated purpose is benign (note-taking), this introduces a significant prompt injection vulnerability. If the agent later reads and interprets this file, a malicious user could craft inputs (e.g., `concept` or `context`) containing agent-specific directives or commands, which would then be written to the file and potentially executed by the agent upon subsequent processing, leading to unauthorized actions or information disclosure. This is a risky capability that could enable an attack without explicit malicious intent in the skill itself.