infsh-cli
Fail
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The installation instructions in SKILL.md and references/authentication.md direct the user to run curl -fsSL https://cli.inference.sh | sh. This downloads a script from a remote URL and executes it immediately in the shell, which is a high-risk pattern for executing unverified code.- [EXTERNAL_DOWNLOADS]: The skill downloads binaries and manifest files from dist.inference.sh as part of its manual installation process described in SKILL.md.- [COMMAND_EXECUTION]: The skill makes extensive use of the belt CLI tool via Bash to interact with the inference.sh platform, including commands to run apps, search for models, and manage tasks.- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its interaction with external model outputs and local files.
- Ingestion points: The agent ingests data from local files (e.g., images, audio, JSON) and AI model outputs when executing belt app run (found in SKILL.md and references/running-apps.md).
- Boundary markers: No delimiters or ignore instructions are provided to separate user data from model instructions.
- Capability inventory: The skill has the capability to read local files, perform network requests, and execute commands through the belt CLI.
- Sanitization: There is no evidence of input validation or output sanitization for the data processed by the tool.
Recommendations
- HIGH: Downloads and executes remote code from: https://cli.inference.sh - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata