hunting-for-suspicious-scheduled-tasks

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONMETADATA_POISONING
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data from Windows event logs (EVTX) and system task exports (CSV/JSON). \n
  • Ingestion points: scripts/agent.py processes EVTX logs via the --evtx argument and CSV task exports via --csv. scripts/process.py processes JSON and CSV event data via the --input argument.\n
  • Boundary markers: There are no explicit boundary markers or instructions to the agent to ignore embedded commands within the processed log data.\n
  • Capability inventory: The skill includes scripts that can read system files and write hunt reports to the local file system.\n
  • Sanitization: The skill lacks sanitization or validation of the ingested data before processing, relying on regex patterns for detection rather than input safety. \n- [DYNAMIC_EXECUTION]: In scripts/agent.py, the hunt_evtx_4698 function uses xml.etree.ElementTree.fromstring(xml_str) to parse XML content extracted from event logs. The standard xml.etree.ElementTree library is vulnerable to XML External Entity (XXE) attacks. If the agent processes a maliciously crafted EVTX file, this could potentially be exploited to perform server-side request forgery (SSRF) or read local files. \n- [EXTERNAL_DOWNLOADS]: The skill references external resources and dependencies for informational purposes. \n
  • The file references/api-reference.md links to the python-evtx library repository on GitHub (https://github.com/williballenthin/python-evtx). \n
  • Multiple files reference Sigma rules and MITRE ATT&CK guidelines hosted on repositories and domains (github.com/SigmaHQ/sigma, attack.mitre.org). These are well-known resources in the cybersecurity domain. \n- [METADATA_POISONING]: There is an inconsistency in the author metadata. The SKILL.md frontmatter identifies the author as 'mahipal', while the LICENSE file attributes the work to 'mukul975'. While this may be a template error, discrepancies in authorship can be a factor in identifying the origin and reliability of a skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 10:52 PM