ezbot-project-research
Fail
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: CRITICALCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/research.pyusessubprocess.runto executecurlcommands. While it uses a list-based argument format, the underlying operation involves spawning a shell process to interact with external networks. \n- [EXTERNAL_DOWNLOADS]: The skill is designed to fetch content from user-provided URLs. Automated scans have confirmed that multiple domains referenced in the skill's documentation, such asptah.tradeandbibiai.fun, are active phishing sites. \n- [REMOTE_CODE_EXECUTION]: The combination of arbitrary external downloads and shell command execution provides a pathway for potential remote code execution if the fetched content or user inputs are not perfectly sanitized. \n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the web usingscripts/inspect_material.py. Evidence chain: \n - Ingestion points: Data is fetched by
scripts/research.pyand parsed byscripts/inspect_material.py(e.g., HTML text, script content). \n - Boundary markers: None identified in the parsing script to separate untrusted web content from agent instructions. \n
- Capability inventory: The skill has access to shell execution (
curl) and file system writes within the designated output directory. \n - Sanitization: The script performs basic whitespace cleaning but lacks robust sanitization against prompt injection payloads embedded in HTML or text. \n- [METADATA_POISONING]: The file
references/agcid-cases.jsonwas flagged as malicious (FileRepMalware) by automated reputation scanners, indicating it may contain malicious payloads or links.
Recommendations
- CRITICAL: 1 file(s) identified as malware by FileRep - DO NOT USE
- AI detected serious security threats
- Contains 13 malicious URL(s) - DO NOT USE
Audit Metadata