truefoundry-ssh-server

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPRIVILEGE_ESCALATIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the tfy CLI and a specialized helper script scripts/tfy-api.sh to interact with the TrueFoundry platform.
  • The tfy-api.sh script serves as an authenticated wrapper for curl, enabling REST API interactions when the primary CLI is unavailable.
  • The script includes input validation to restrict HTTP methods and implements checks to prevent path traversal attacks in API paths.
  • [EXTERNAL_DOWNLOADS]: The skill manages the installation of vendor-specific tools and container images from trusted registries.
  • It automates the installation of the truefoundry CLI using pip or uv with pinned versions for stability.
  • Development environments are provisioned using container images hosted on public.ecr.aws and tfy.jfrog.io, which are vendor-controlled repositories.
  • Networking utilities like proxytunnel are recommended for installation via official package managers such as Homebrew.
  • [PRIVILEGE_ESCALATION]: Certain setup tasks require elevated privileges on the local machine and within the cloud environment.
  • The instructions specify the use of sudo for installing networking tools on the user's local machine, accompanied by security warnings and requests for user confirmation.
  • Deployment templates for Docker images utilize USER root to facilitate the installation of system-level dependencies during the build phase.
  • [CREDENTIALS_UNSAFE]: The skill handles authentication data required for remote access and API authorization.
  • It reads the user's public SSH key from ~/.ssh/id_rsa.pub to configure authorization for the deployed SSH server.
  • API authentication is managed through TFY_API_KEY, which is read from environment variables or a local .env file, avoiding hardcoded secrets.
  • The skill promotes the use of tfy-secret:// URIs for managing sensitive application credentials within deployment manifests.
  • [INDIRECT_PROMPT_INJECTION]: The skill has a surface for processing external data that could be influenced by untrusted sources.
  • Ingestion points: Ingests code from remote Git repositories defined in the repo_url field for automated build and deployment processes.
  • Boundary markers: Implements strict operational rules requiring the agent to obtain explicit user confirmation before selecting workspaces or applying public SSH keys.
  • Capability inventory: Includes capabilities for network communication via tfy-api.sh, file system interaction for manifest generation, and remote code execution through the platform's build system.
  • Sanitization: While no automated sanitization of repository content is described, the skill relies on manual user review of manifests and dry-run previews.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 03:58 PM