ambit-cli

Pass

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the @cardelli/ambit CLI via npx to manage private networks and perform application deployments.
  • [EXTERNAL_DOWNLOADS]: Fetches application templates from the vendor's GitHub repository (ToxicPine/ambit-templates) or user-provided repositories during the deploy command execution.
  • [REMOTE_CODE_EXECUTION]: Deploys application code from remote templates to Fly.io infrastructure; the template content is executed within the user's private network environment.
  • [CREDENTIALS_UNSAFE]: Requires a Tailscale API access token (tskey-api-...) to manage network ACLs and router authentication. The documentation states that this key is used locally and is not sent to the routers.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface through the --template flag, which allows fetching and deploying code from arbitrary GitHub repositories.
  • Ingestion points: SKILL.md describes the --template argument for the deployment command.
  • Boundary markers: Absent; no explicit instructions are provided to the agent to ignore instructions embedded within retrieved templates.
  • Capability inventory: npx @cardelli/ambit deploy executes code deployment and configuration on Fly.io infrastructure.
  • Sanitization: Documentation mentions pre-flight scans for dangerous settings in configuration files, but does not specify protection against general malicious code in templates.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 9, 2026, 11:45 PM