Folder Pilot — agentic threat model

Not certain from the listing — The specific foundation models used for OCR, tagging, and chat are not disclosed. Potential threats include adversarial prompt injection via uploaded documents (e.g., indirect prompt injection in OCR text) leading to unauthorized file sharing or data exfiltration.

The agent performs OCR, automatic tagging, and RAG ('chat with your files'). Key threats include data poisoning via malicious files uploaded for OCR, embedding inversion, and unauthorized data exfiltration if sharing controls are bypassed.

Not certain from the listing — The underlying agent framework is unspecified. Threats include insecure tool integration where file management commands (delete, move, share) could be abused via prompt injection or malicious file names.

Not certain from the listing — As an open-source tool, deployment could be local or cloud-hosted. Risks include local privilege escalation, directory traversal via the file management interface, and insecure sharing endpoints exposing local files to the network.

Not certain from the listing — No evaluation, guardrails, or observability features are mentioned. Lack of monitoring could allow silent data exfiltration or unauthorized file sharing to go undetected.

Not certain from the listing — While 'file and folder sharing' is featured, the authorization model and access controls are not detailed. Weak access controls could lead to unauthorized sharing of sensitive OCR-extracted data.

Not certain from the listing — There is no indication of multi-agent orchestration or ecosystem integration. The primary risk is limited to standalone file management and sharing.