kdocs

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill instructs the agent to accept a user-provided Token and to construct/run a mcporter command that embeds "Authorization=Bearer " (a direct command-line/header value), which requires including the secret verbatim and thus presents an exfiltration risk despite admonitions not to echo it.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the agent to fetch and ingest arbitrary public webpages via the scrape_url flow (and to consume external "references" from aippt.deep_research) and then use that content in downstream analysis and automated actions (summarization, outline/PPT generation, file creation), so untrusted third‑party content can materially influence the agent’s decisions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill instructs the agent to register and call an external MCP service at "https://mcp-center.wps.cn/skill_hub/mcp" (mcporter config add) and to run remote update checks whose returned "instruction" field is acted on at runtime, meaning this URL is used during runtime and can deliver instructions that directly control agent actions.