capgo-cli-usage

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill instructs the agent to use npx @capgo/cli@latest to perform operations. This command downloads and executes the vendor's CLI tool from the npm registry. The skill correctly advises pinning versions in CI environments for better security.\n- [COMMAND_EXECUTION]: The agent is directed to run various shell commands to manage apps, bundles, and channels. These instructions include a safety requirement to seek user confirmation before performing destructive or state-changing actions.\n- [CREDENTIALS_UNSAFE]: The skill manages the lifecycle of Capgo API tokens and environment variables like CAPGO_TOKEN. It provides specific guidance to avoid exposing these credentials in chat logs or shell history and identifies vendor-specific configuration paths like ~/.capgo.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests user input to parameterize CLI commands, creating an indirect injection surface. This is mitigated by explicit boundary instructions requiring human confirmation for executed actions and instructions for handling sensitive data that may appear in the conversation flow.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 12:04 AM