pi-coding-agent-sdk
Pass
Audited by Gen Agent Trust Hub on Mar 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a knowledge base for the Pi Coding Agent SDK, facilitating legitimate software development and integration tasks.
- [DATA_EXFILTRATION]: Analysis identified mentions of sensitive credential storage files (e.g.,
auth.json,models.json) and configuration paths. These references are part of the SDK's documentation explaining how it manages authentication and settings, with no instructions to exfiltrate this data. - [COMMAND_EXECUTION]: The skill documents the use of powerful SDK tools like
bashToolandeditTool. These are presented as standard components for a coding agent's operation, and the instructions focus on their correct programmatic implementation rather than malicious usage. - [PROMPT_INJECTION]: The documentation describes how the SDK ingests repository-level context files (
AGENTS.md, prompts, and skills). This identifies the inherent indirect prompt injection surface common to coding agents, but the skill itself does not contain malicious instructions or bypass attempts.
Audit Metadata