PearAI — agentic threat model
PearAI operates with high local privileges as a VSCode fork, integrating powerful code-generation and memory tools like Aider and Mem0. Its primary risk is passive prompt injection from untrusted codebases, which could lead to unauthorized local file modification or command execution on the developer's machine.
OWASP AIVSS score rationale
| Autonomy of Action | 0.50 | |
| Goal-Driven Planning | 0.60 | |
| Self-Modification | 0.20 | |
| Dynamic Tool Use | 0.80 | |
| Persistent Memory | 0.70 | |
| Contextual Awareness | 0.80 | |
| Dynamic Identity | 0.10 | |
| Multi-Agent Interactions | 0.50 | |
| Non-Determinism | 0.70 | |
| Opacity & Reflexivity | 0.50 |
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 — The specific LLMs used are not detailed, but as an open-source IDE fork, it likely supports multiple commercial and local models. The primary threat is adversarial prompt injection via open files (passive injection) causing the model to generate malicious code or commands.
PearAI utilizes the developer's local codebase as context, integrates Mem0 for persistent memory, and uses Perplexity for web search. Threats include codebase poisoning (where malicious files in a repository manipulate the RAG context) and memory poisoning within Mem0.
The orchestration relies on integrations like Aider and Continue. Insecure tool integration is a major threat here; if the agent framework executes file writes or terminal commands based on untrusted LLM outputs, it can lead to local code execution.
As a local VSCode fork, PearAI runs directly on the developer's workstation. A compromise of the editor or its integrated agents inherits the developer's local user privileges, risking exposure of local SSH keys, environment variables, and source code.
Not certain from the listing — There is no mention of built-in guardrails, output sanitization, or safety monitoring to detect and block malicious code generation or unauthorized file system access.
Not certain from the listing — Compliance controls, telemetry opt-out policies, and enterprise security configurations are not detailed in the public directory listing.
PearAI acts as an ecosystem hub integrating multiple external services (Supermaven, Perplexity, Mem0, Aider). This creates a complex supply chain threat where a compromise or data leakage in any of these third-party services directly impacts the developer's IDE environment.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).
These scores are auto-generated from public information (the agent's own listing, docs, and repository) using the canonical OWASP AIVSS formula and the MAESTRO framework — an estimate for guidance, not a penetration test, audit, or certification. See the scoring methodology. Are you the vendor? Factual corrections are free.