starknet-agentic-skills

Fail

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: CRITICALCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: Several instruction files guide users to install software by piping remote scripts directly into a shell (e.g., curl | bash). This pattern bypasses local validation and relies entirely on the integrity of the remote server. Evidence found in:
  • skills/cairo-deploy/SKILL.md: Fetches and executes from foundry-rs/starknet-foundry.
  • skills/controller-cli/SKILL.md: Fetches and executes from cartridge-gg/controller-cli.
  • skills/huginn-onboard/SKILL.md: Fetches and executes from welttowelt/daydreams.
  • [EXTERNAL_DOWNLOADS]: The documentation and example configurations reference https://starknet-faucet.vercel.app/ for obtaining testnet funds. This domain has been blacklisted by automated security scanners (e.g., URLite).
  • [COMMAND_EXECUTION]: A significant number of internal utility and quality-assurance scripts execute shell commands via subprocess.run(). This includes complex operations such as cloning and scanning external repositories. Key files include:
  • scripts/quality/audit_local_repo.py
  • scripts/quality/scan_external_repos.py (clones external repos via CLI)
  • scripts/quality/benchmark_cairo_auditor.py
  • [DATA_EXFILTRATION]: The starknet-anonymous-wallet skill contains scripts (_keys.js) that are designed to read private keys from sensitive local paths such as ~/.openclaw/secrets/starknet.
Recommendations
  • CRITICAL: 1 file(s) identified as malware by FileRep - DO NOT USE
  • AI detected serious security threats
  • Contains 4 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Aug 4, 2026, 04:03 AM
Security Audit — agent-trust-hub — starknet-agentic-skills