paper7-research

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly searches and fetches papers from arXiv via the "paper7" tool (see "Search for Papers" and "Build Knowledge Base" / example commands in SKILL.md) and then reads and synthesizes those public papers to drive recommendations and next actions, so untrusted third-party content can materially influence agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches arXiv papers at runtime (e.g., via paper7 get and arXiv URLs such as https://arxiv.org/abs/XXXX.XXXXX), and those fetched documents are injected into the agent's knowledge base and directly drive synthesized prompts/output, so this external content controls the agent at runtime.