thesis-workflow-agent

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill explicitly instructs the agent to ask users to send API keys in the current session and to inject those keys into local configuration, which requires the agent to handle secret values and creates a high risk the LLM will include them verbatim in generated commands or config instructions.

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to configure and use third‑party services (Metaso MCP via https://metaso.cn for literature search and Paperyy via https://www.paperyy.com/ for AIGC/plagiarism reports) and to read/parse those external search results and downloaded reports (see Phase 0 and Phase 7 in SKILL.md and references/report-handling.md), and then use those untrusted third‑party contents to drive optimizations and DOCX edits—i.e., the agent ingests and acts on public third‑party content as part of its workflow.