skills/whytryharder/skills/infsh-cli/Gen Agent Trust Hub

infsh-cli

Fail

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill instructs the agent/user to install the core CLI tool by piping a remote script directly into the shell: curl -fsSL https://cli.inference.sh | sh. This is a high-risk pattern that executes unverified code from an external server without manual review or checksum verification during the primary installation path.
  • [DATA_EXFILTRATION]: The belt CLI tool contains an 'automatic upload' feature. If a local file path is provided in an input parameter instead of a URL, the tool reads the file from the local disk and uploads it to the inference.sh cloud. This behavior creates a significant data exfiltration risk if the agent is manipulated into processing sensitive paths such as ~/.ssh/id_rsa, ~/.aws/credentials, or .env files.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted data (prompts and file paths) and interpolates them into commands executed by the belt CLI.
  • Ingestion points: Untrusted data enters the context through input JSON strings in SKILL.md and references/running-apps.md used with the belt app run command.
  • Boundary markers: None identified in the provided instructions to separate instructions from untrusted data.
  • Capability inventory: The skill uses the Bash tool to run the belt CLI, which has capabilities for network access, file reading, file uploading, and cloud deployment.
  • Sanitization: No evidence of input validation or sanitization before passing parameters to the CLI.
  • [COMMAND_EXECUTION]: The skill facilitates the execution of the belt CLI, which provides access to a wide range of external services and AI models, including capabilities for Twitter automation, web searching, and executing remote 'apps' that could perform varied tasks.
  • [PERSISTENCE_MECHANISMS]: The CLI reference documentation in references/cli-reference.md includes commands to write shell completion scripts to system-protected or configuration directories like /etc/bash_completion.d/ and ~/.config/fish/completions/. While intended for usability, modifying these paths is a common persistence technique.
Recommendations
  • HIGH: Downloads and executes remote code from: https://cli.inference.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Aug 6, 2026, 05:03 AM
Security Audit — agent-trust-hub — infsh-cli