flyio-cli-public
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
flyctlCLI to perform management tasks. Key capabilities include executing remote commands viafly ssh console -C "...", deploying code, and modifying environment secrets usingfly secrets set. These operations grant the agent significant control over the user's remote application environment. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data which could potentially contain malicious instructions.
- Ingestion points: Data is ingested from application logs (
fly logs), application status outputs, and project configuration files (fly.toml). - Boundary markers: The instructions do not define clear delimiters or "ignore" directives to help the agent distinguish between data content and instructions.
- Capability inventory: The agent has the capability to execute shell commands on remote Fly.io machines, manage application secrets, and change deployment configurations.
- Sanitization: There is no evidence of sanitization or strict schema validation for the data ingested from logs or configuration files.
Audit Metadata