truefoundry-notebooks
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Automatically installs the TrueFoundry CLI (
truefoundry) and thepydanticlibrary from official package registries to ensure environment compatibility.\n - Fetches container images from official TrueFoundry repositories on well-known registries such as Amazon ECR Public Gallery, GitHub Container Registry, and NVIDIA Container Registry.\n- [COMMAND_EXECUTION]: Executes the vendor's command-line tool (
tfy) and a distributed API helper script (tfy-api.sh) to manage notebook services and discover workspace resources.\n- [INDIRECT_PROMPT_INJECTION]: The skill maintains a standard attack surface for processing untrusted data during the deployment workflow.\n - Ingestion points: Collects user input for notebook identification, resource limits (CPU, memory, GPU), and auto-shutdown cull timeouts.\n
- Boundary markers: Deployment manifests are generated by interpolating variables into YAML templates and shell heredocs.\n
- Capability inventory: Capability to create cloud resources via
tfy applyand perform authenticated network operations viacurl(encapsulated intfy-api.sh).\n - Sanitization: Relies on manifest schema enforcement; no explicit sanitization of user-provided strings is implemented within the skill's local logic prior to manifest generation.
Audit Metadata