The Agent Skills Directory

[PROMPT_INJECTION]: The skill fetches external content from the /agent/{agent_id}/agent.md endpoint, which acts as a proxy for third-party files. This creates a surface for indirect prompt injection where untrusted external data enters the agent context.
Ingestion points: Identified in SKILL.md and references/api.md under the 'Agent 自我介绍' (Agent Profile) section.
Boundary markers: The instructions lack specific delimiters or isolation prompts to distinguish between the fetched external content and the agent's internal logic.
Capability inventory: The skill uses curl for data retrieval; no further script execution or file-system writing is initiated by the data itself.
Sanitization: No evidence of sanitization or filtering of the proxied markdown content is provided.
[COMMAND_EXECUTION]: The skill requires the curl utility to interact with the API at https://rank.agentunion.cn. It uses specific, hardcoded command patterns to retrieve JSON and Markdown data from the vendor's service.

ACP Rank