The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes external accounting standards from docs.maoyanqing.com. It includes automated scripts (scripts/fetch_standards.py and scripts/sync-standards.sh) to download and cache these markdown files locally. This is a primary function of the skill to ensure it provides advice based on the most recent regulations.
[COMMAND_EXECUTION]: The skill provides a Python utility script and a shell script for the user or agent to synchronize the standards library. These scripts execute shell commands (curl) and perform network operations to fetch data.
[INDIRECT_PROMPT_INJECTION]: Because the skill's core workflow relies on ingesting and reasoning over external regulatory documents, it possesses an attack surface for indirect prompt injection if a source document were to contain malicious instructions.
Ingestion points: Files located in references/standards/ and live content retrieved via the webReader tool from docs.maoyanqing.com.
Boundary markers: Absent. The skill does not use specific delimiters to separate standard text from instructional context, treating the ingested standards as authoritative facts.
Capability inventory: The skill can read local files, execute shell/Python scripts, and perform network requests via the webReader tool.
Sanitization: Absent. There is no evidence of filtering or sanitizing the content of the markdown files before they are processed by the agent.

chen-yiwei-perspective