predict-worknet

Fail

Audited by Gen Agent Trust Hub on May 6, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill includes instructions to install its core component by piping a remote shell script directly into the system shell (curl ... | sh). This script is hosted on the author's GitHub repository.
  • [COMMAND_EXECUTION]: The skill relies on several external binaries, including predict-agent, awp-wallet, and openclaw, which it executes as subprocesses to manage its autonomous prediction loop and wallet operations.
  • [EXTERNAL_DOWNLOADS]: The skill fetches pre-compiled binaries and source code from the author's GitHub repositories during its setup and registration phases.
  • [PROMPT_INJECTION]: The prediction loop retrieves natural language "challenge" prompts from the prediction server and includes them in the agent's prompt during execution. This provides a surface for indirect prompt injection if the external server's content is compromised.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/awp-worknet/prediction-skill/main/install.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
May 6, 2026, 02:03 AM