SciSummary — agentic threat model
SciSummary presents a low-to-moderate risk profile, primarily centered around data privacy and indirect prompt injection. The main threats involve malicious document uploads designed to hijack the summarization LLM or exploit the PDF parsing infrastructure.
OWASP AIVSS score rationale
| Autonomy of Action | 0.20 | |
| Goal-Driven Planning | 0.30 | |
| Self-Modification | 0.10 | |
| Dynamic Tool Use | 0.40 | |
| Persistent Memory | 0.50 | |
| Contextual Awareness | 0.50 | |
| Dynamic Identity | 0.10 | |
| Multi-Agent Interactions | 0.10 | |
| Non-Determinism | 0.50 | |
| Opacity & Reflexivity | 0.40 |
Scored with the canonical OWASP AIVSS formula (AIVSS calculator reference); agentic risk factors estimated from the agent’s described capabilities.
MAESTRO 7-layer threat model
Per-layer threats for this agent. Layers tagged “not certain from listing” are general, caveated commentary where the public description didn’t pin that layer.
Not certain from the listing — likely utilizes commercial LLMs (such as GPT or Claude) for summarization. The primary threat is indirect prompt injection, where instructions embedded in uploaded research papers hijack the model's behavior during summarization or chat.
The agent ingests user-uploaded documents, extracts references, and queries a 200M paper database. Key threats include data exfiltration of proprietary pre-publication papers, vector database poisoning, and unauthorized access to other users' private libraries.
Not certain from the listing — likely uses a standard RAG framework (e.g., LangChain) to orchestrate document parsing, vector search, and LLM querying. Threats include insecure document parsing (e.g., PDF parser exploits) and SSRF if importing references via arbitrary external URLs.
Not certain from the listing — likely hosted on standard cloud infrastructure with a vector database. Threats include container compromise during resource-intensive PDF parsing and unauthorized API access to the document storage bucket.
Not certain from the listing — no details on evaluation or guardrails. Gaps could lead to undetected hallucinations or toxic outputs when summarizing controversial or adversarial papers.
Not certain from the listing — requires standard web authentication for library access. Compliance risks include handling copyrighted material, proprietary pre-prints, and ensuring GDPR/CCPA compliance for user-uploaded documents.
Not certain from the listing — does not appear to interact with external agent marketplaces or third-party autonomous agents, limiting ecosystem risks to standard API integrations.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).