cti-domain-research
Pass
Audited by Gen Agent Trust Hub on Apr 29, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands within the
install.shscript to copy configuration files and slash commands to the user's local.claudedirectory. It also invokes a local Node.js script to process research data during the NotebookLM export workflow. - [EXTERNAL_DOWNLOADS]: The documentation and installation scripts reference an external GitHub repository (
github.com/Security-Phoenix-demo/security-skills-claude-code) to download an optional connector plugin for NotebookLM integration. This resource belongs to the skill's authoring organization. - [DATA_EXFILTRATION]: When the
--notebooklmflag is enabled, the skill transmits collected research findings and source URLs to the external NotebookLM service using user-provided API identifiers. This is the intended behavior of the research-sharing feature. - [PROMPT_INJECTION]: The skill processes untrusted data from various external web sources (security blogs and advisories). While it functions as a research aggregator with a broad ingestion surface, the skill instructions focus on structured formatting and do not contain patterns attempting to bypass agent safety filters.
Audit Metadata