ktx-ai-data-agents
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill installs and uses the
@kaelio/ktxpackage from npm and references thektx_slPython library. - [COMMAND_EXECUTION]: The skill requires the execution of several shell commands for setup and operation, including
npm install,npx, and CLI tools for context ingestion and starting an MCP server. - [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests unstructured data from external sources to build the AI agent's context.
- Ingestion points: The
ktx ingestcommand processes data from Notion pages and dbt documentation files (SKILL.md). - Boundary markers: The system uses a structured semantic layer defined in YAML to guide agent queries, but does not specify delimiters for prose ingested from wiki sources.
- Capability inventory: Provides tools for searching the semantic layer, searching business knowledge (wiki), describing entities, and listing database connections.
- Sanitization: No explicit sanitization or filtering of external content is documented before it is provided as context to the agent.
Audit Metadata