skills/arthjean/skills/neon-cli/Gen Agent Trust Hub

neon-cli

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses bunx neonctl@latest to dynamically fetch and execute the Neon CLI tool. This targets a well-known service repository on a public registry.- [COMMAND_EXECUTION]: The skill executes shell scripts that wrap neonctl and psql commands to perform database operations, schema inspections, and management tasks.- [INDIRECT_PROMPT_INJECTION]: The skill processes SQL queries and Neon Management API calls based on instructions or data. It handles database resources such as projects and branches.
  • Ingestion points: SQL strings passed to neon-sql.sh, neon-tx.sh, and neon-explain.sh; JSON request bodies and fields passed to neonctl api as documented in references/management-api.md.
  • Boundary markers: The instructions recommend using psql variables (-v) and heredocs with quoted delimiters (<<'SQL') to prevent shell interpolation.
  • Capability inventory: Execution of SQL (DDL/DML) via psql, file-based transactions, management of Neon projects and branches, and network API requests via neonctl.
  • Sanitization: Instructions explicitly state to treat table, schema, role, and database identifiers as untrusted input and to avoid concatenating untrusted values into SQL strings.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 08:54 PM