pgpm-extensions
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill utilizes
pgpm installto download modules from the npm registry under the@pgpmscope (e.g.,@pgpm/faker,@pgpm/base32). These are standard vendor resources for the pgpm ecosystem. - [COMMAND_EXECUTION]: The agent is directed to execute CLI commands such as
pgpm extension,pgpm install, andpgpm upgrade-modulesto manage the PostgreSQL environment. - [PROMPT_INJECTION]: The skill describes an indirect injection surface through the management of
.controlfiles for dependency resolution. - Ingestion points: The agent reads dependency lists from the
requiresfield in module.controlfiles. - Boundary markers: No explicit markers are defined to help the agent distinguish between data and instructions within the
.controlfile content. - Capability inventory: The agent has the capability to install and deploy code using the pgpm CLI based on the contents of these configuration files.
- Sanitization: There are no instructions for validating or sanitizing the strings read from the
.controlfiles. - [SAFE]: No obfuscation, data exfiltration, or malicious persistence mechanisms were identified. The 'Critical Rule' regarding
CREATE EXTENSIONis a functional constraint to ensure database consistency.
Audit Metadata