expanso-s3-to-postgres

Pass

Audited by Gen Agent Trust Hub on Jul 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) as it processes data from external, untrusted sources.
  • Ingestion points: The pipeline.yaml file defines an aws_s3 input that reads objects from an S3 bucket.
  • Boundary markers: None. The skill does not implement delimiters or instructions to ignore embedded prompts within the ingested JSON data.
  • Capability inventory: The skill has the capability to write data to a PostgreSQL database (sql_insert in pipeline.yaml) and execute shell commands via run.sh to trigger the expanso binary.
  • Sanitization: The skill uses Bloblang mapping for structural transformation, but does not perform content-level sanitization or filtering to prevent malicious instructions from being passed to downstream consumers of the PostgreSQL data.
  • [COMMAND_EXECUTION]: The run.sh script executes the expanso binary with user-supplied arguments ($@). While expanso is a tool associated with the vendor, passing unvalidated shell arguments to a binary can lead to command injection if the agent or user provides malicious flags.
  • [DATA_EXFILTRATION]: The skill performs network-based data transfers. It reads from AWS S3 and writes to a PostgreSQL instance. While the domains are controlled via environment variables (S3_BUCKET, POSTGRES_DSN), the pattern involves moving potentially sensitive data across network boundaries.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 26, 2026, 05:45 AM
Security Audit — agent-trust-hub — expanso-s3-to-postgres