data-programs
Warn
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: MEDIUMREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill facilitates the creation, storage, and execution of JavaScript scripts (run-code) using actions like save-data-program and run-data-program. These scripts are executed within a sandbox that provides access to internal platform functions and network capabilities.
- [DATA_EXFILTRATION]: Scripts have access to the webFetch global, which allows them to make external network requests. Because these scripts can also access sensitive data via providerFetch and appAction using the current viewer's credentials, a malicious script could potentially exfiltrate sensitive data to an external server when executed by other users in the organization.
- [PROMPT_INJECTION]: The skill processes data from untrusted external sources, creating a surface for indirect prompt injection.
- Ingestion points: Data is ingested via providerFetch, webFetch, and workspace* file access as described in SKILL.md.
- Boundary markers: There are no instructions for the agent to use specific delimiters or ignore instructions embedded in the processed data content.
- Capability inventory: The environment includes webFetch for network access, save-data-program for code persistence, and emit for data output to the UI.
- Sanitization: No content-level sanitization or filtering is described for the fetched data; limits are restricted to row counts and result sizes.
Audit Metadata