The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data from multiple sources including user-provided Markdown, HTML, and text files, as well as remote URLs and comments retrieved via note lookup commands. While the skill mandates a conversion process to a specific JSON Abstract Syntax Tree (AST) schema and includes logic to strip unsupported tags, these ingestion points represent an attack surface for indirect prompt injection. The skill possesses capabilities to modify account state (create/edit notes, set privacy) and upload files, which could potentially be manipulated by malicious instructions embedded within the processed content. Evidence includes the use of mocli note info --show-comment and content conversion logic for external files.
[COMMAND_EXECUTION]: The skill utilizes the mocli command-line tool to perform operations such as note creation, editing, and searching. The instructions guide the agent in executing these commands via subprocesses, including handling input via stdin or temporary files.
[EXTERNAL_DOWNLOADS]: The skill supports uploading local assets or fetching content from remote URLs to the Mowen Note service using mocli misc upload. These operations target the vendor's official domain (mowen.cn) and are documented as part of the core functionality for resource management within notes.

mo-note