endor-agent-kit-setup
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local configuration files (
~/.endorctl/config.yaml) and environment variables (ENDOR_NAMESPACE) to determine the environment state. This creates a potential surface for indirect prompt injection if an attacker can manipulate these local configuration sources to influence the agent's logic. Evidence: 1. Ingestion points:~/.endorctl/config.yaml,ENDOR_NAMESPACE. 2. Boundary markers: The skill uses a structured 'Readiness Report' to separate and report findings. 3. Capability inventory: The skill can write host-specific MCP configuration files and facilitate the installation of theendorctlbinary. 4. Sanitization: Instructions mandate 'safely parse' logic for non-secret fields and explicitly forbid printing or asking for secret values. - [EXTERNAL_DOWNLOADS]: The skill provides instructions for installing the
endorctlCLI tool via package managers or direct binary download with checksum verification, as well as installing plugins from theendorlabs/ai-pluginsmarketplace. These resources originate from the verified vendor. - [COMMAND_EXECUTION]: The skill uses various command-line tools for diagnostic purposes, including
gh auth status,git, andendorctl. It also describes configuring an MCP server usingnpx -y endorctl ai-tools mcp-server.
Audit Metadata