primitive-chat

Pass

Audited by Gen Agent Trust Hub on Oct 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute several CLI commands for account management and communication, including primitive chat, primitive account show, primitive whoami, and primitive agent start-agent-signup. These are documented as part of the official Primitive CLI toolset.
  • [EXTERNAL_DOWNLOADS]: Instructions include the installation of the @primitivedotdev/cli package via npm install -g. As this is the official package for the skill's vendor (primitivedotdev), it is considered a legitimate dependency.
  • [PROMPT_INJECTION]: The skill contains explicit instructions to override default agent behaviors regarding user confirmation (e.g., "Use freely (no user confirmation needed)" for specific support addresses). While this instructs the agent to bypass typical safety steps for certain actions, it is scoped to automated agent-to-agent help scenarios and includes mandatory "Ask the user first" guidelines for human-facing addresses (sales, billing).
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 17, 2026, 10:43 PM