accounts
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the bun runtime to execute locally bundled TypeScript scripts, such as metering-engine.ts, reconcile-gateways.ts, and pricing-calculator.ts, to perform financial calculations and generate reports based on processed data.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external financial records and usage telemetry files, which serves as a data ingestion surface.
- Ingestion points: Usage events and settlement records are ingested via file paths provided as arguments to the metering-engine.ts and reconcile-gateways.ts scripts.
- Boundary markers: The skill incorporates the 'Vibeguard Protocol' within SKILL.md to instruct the agent to redact sensitive credentials like Stripe keys or banking details. It also follows the 'OpenAccountants' procedure for structured data classification.
- Capability inventory: The skill is configured with tools for file manipulation and command execution (bash, run_command, write_to_file). The included scripts perform filesystem operations to record compute token usage and generate pricing schedules.
- Sanitization: Input data is parsed as JSON and handled using typed interfaces in TypeScript to ensure that incoming data conforms to the expected financial schemas before it is used in calculations.
Audit Metadata