spacetimedb-tables
Pass
Audited by Gen Agent Trust Hub on May 7, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill instructions do not contain any patterns attempting to override agent behavior, bypass safety filters, or extract system prompts. The language is purely instructional and focused on database schema design.
- [DATA_EXFILTRATION]: No sensitive local file access or hardcoded credentials were found. While the documentation in
tables-file-storage.mdprovides examples for uploading data to external storage (e.g., AWS S3), these are legitimate use cases for the database platform being documented. The code uses parameters for buckets and regions and specifically advises on proper authentication practices. - [EXTERNAL_DOWNLOADS]: The skill does not perform any unauthorized remote code downloads or script executions. It references official SpacetimeDB libraries and GitHub issue trackers for technical context.
- [INDIRECT_PROMPT_INJECTION]: The skill documents ingestion points for untrusted data, such as user profiles and file uploads in
tables-file-storage.mdandtables-access-permissions.md. While this establishes an attack surface where an agent might process malicious content stored in the database, the skill explicitly provides guidance on using 'Private' tables for sensitive data and project-specific access controls to mitigate these risks. - [COMMAND_EXECUTION]: No dangerous shell command execution or unauthorized subprocess spawning was detected. The code snippets provided are language-specific API calls for the SpacetimeDB environment.
Audit Metadata