Plandex AI Coding Agent for Large Projects
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The Plandex agent is designed to process extensive project context (up to 2M tokens), which creates a surface for indirect prompt injection if the ingested source code contains malicious instructions.
- Ingestion points: Local project files and source code loaded into the agent's context via the REPL or file-loading commands.
- Boundary markers: The skill file does not define specific delimiters or instructions to the agent to ignore embedded prompts within the project files it processes.
- Capability inventory: The agent is a terminal-based tool capable of suggesting and applying complex code changes across large projects.
- Sanitization: The tool utilizes a "diff sandbox" intended to provide a layer of human review before changes are finalized, mitigating the impact of potentially unintended modifications.
Audit Metadata