lakebase-scm-workflows
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONCREDENTIALS_UNSAFE
Full Analysis
- [COMMAND_EXECUTION]: Orchestrates development workflows by executing shell commands including
git,databricksCLI, and specializedlakebase-*binaries for project setup and branch synchronization. - [COMMAND_EXECUTION]: Installs git hooks (
post-checkout,post-merge,pre-push,prepare-commit-msg) into the local repository that automatically trigger synchronization scripts during standard git operations. - [EXTERNAL_DOWNLOADS]: Configures dependencies and fetches the
lakebase-app-dev-kitfrom vendor-owned GitHub repositories and official npm registries. - [DATA_EXFILTRATION]: Manages project configuration and credentials by writing
DATABASE_URLand branch metadata to local.envfiles and interacting with Databricks and GitHub APIs for token resolution. - [CREDENTIALS_UNSAFE]: Includes documentation with example DSN strings containing placeholder authentication tokens (JWTs) used strictly for instructional purposes.
- [SAFE]: Evaluated the surface for indirect prompt injection:
- Ingestion points: Data from PR comments, PR file contents, and git branch names.
- Boundary markers: Reliance on agent context without explicit delimiters for all inputs.
- Capability inventory: Shell command execution, filesystem modifications, and network access.
- Sanitization: Branch names are sanitized to restricted alphanumeric formats to prevent injection into commands.
Audit Metadata