The Agent Skills Directory

[DATA_EXFILTRATION]: An automated antivirus scanner detected a high-risk infection (MD:HttpRequest-inf) in the file references/log-sources-ingestion.md. This signature is associated with logic that performs unauthorized or malicious network requests, potentially for data harvesting.
[EXTERNAL_DOWNLOADS]: Multiple reference files, including architecture-patterns.md and deployment-pipelines.md, contain repeated code templates for performing high-concurrency network fetches using aiohttp.ClientSession. In the context of the confirmed AV detection, these templates pose a high risk for large-scale data exfiltration.
[PROMPT_INJECTION]: The skill employs a concealment tactic involving extreme redundancy. Eight separate reference files consist of 150 identical or near-identical sections of dense technical boilerplate. This is a known adversarial technique designed to bury malicious content and overwhelm both human reviewers and automated security scanners.

siem-engineering