pscale-import-d1
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of Cloudflare D1 SQL exports, creating a potential vector for indirect prompt injection if the content of these exports is designed to exploit the agent's parsing or command-generation logic.
- Ingestion points: The skill instructions focus on processing external files provided via the
--inputflag (e.g.,./d1-export.sql) inSKILL.md. - Boundary markers: The documentation includes explicit warnings to "treat the export and generated DDL as untrusted input" and mandates human inspection of the conversion results.
- Capability inventory: The skill leverages subprocess execution of
pscale,wrangler, andpgloaderto perform data extraction, linting, conversion, and loading operations. - Sanitization: The workflow relies on a mandatory
pscale import d1 lintstep to identify incompatibilities and block execution on errors, providing a layer of technical validation before data is loaded. - [COMMAND_EXECUTION]: The skill coordinates the execution of multiple command-line interfaces including
pscale,wrangler, andpgloaderto manage database states and perform schema transformations.
Audit Metadata