The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the open shell command to automatically launch a generated HTML report file in the user's default web browser upon completion of the decision process.
[EXTERNAL_DOWNLOADS]: Fetches the skill definition metadata and configuration from the vendor's public GitHub repository (github.com/Backtthefuture/huangshu) using curl or npx during installation.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted user input (business topics and scenarios) which is then rendered into a local HTML file. Because the agent is instructed to open this file in a browser, malicious instructions embedded in the initial input could potentially execute scripts if not properly handled by the agent's internal rendering logic.
Ingestion points: User-provided business issues captured in SKILL.md (Phase 0: 议题接收).
Boundary markers: Absent; there are no instructions to delimit or ignore instructions within the user-provided data.
Capability inventory: The skill utilizes the Write tool to create local files and the agent's shell capability to execute the open command.
Sanitization: Absent; the instructions for Phase 6 (HTML Visual Report) define the visual design and structure but do not specify sanitization or escaping requirements for the interpolated user content.

advisory-board