pp-jobber

Pass

Audited by Gen Agent Trust Hub on Jul 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the jobber-pp-cli binary using npx from the @mvanhorn NPM scope or via go install from the github.com/mvanhorn/printing-press-library repository. These locations are associated with the recognized software vendor.
  • [COMMAND_EXECUTION]: The skill requires the execution of the jobber-pp-cli tool to perform operations such as AR aging analysis, invoice tracing, and database snapshots. It also involves shell commands for verification and installation (which, npx, go).
  • [DATA_EXFILTRATION]: The CLI includes a --deliver webhook:<url> feature that allows the output of commands to be POSTed to a user-specified URL. Additionally, the feedback command can transmit data to a remote endpoint if the JOBBER_FEEDBACK_ENDPOINT environment variable is configured. While these are documented features, they represent capabilities for transmitting data externally.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external Jobber API connections and processes it into a local SQLite database for full-text search and SQL querying.
  • Ingestion points: Data is synced from the Jobber API into a local SQLite store (SKILL.md).
  • Boundary markers: The skill documentation specifies that ad-hoc SQL is restricted to read-only mode (mode=ro) and the CLI itself is read-only by design.
  • Capability inventory: The skill uses subprocess calls for CLI operations, performs network operations via webhooks, and writes to local files or named pipes.
  • Sanitization: The skill emphasizes the use of a local SQLite store to isolate analysis from the live API, though explicit sanitization of the Jobber data content is not detailed.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 15, 2026, 10:14 PM
Security Audit — agent-trust-hub — pp-jobber