The automatic agent setup matches the warning-only agent extension lifecycle risk: it installs the package's own skills and registers its own MCP server during postinstall. Runtime telemetry can include the operator's email, but uses a user-configured collector, so the evidence does not establish malicious exfiltration.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
Package installation triggers the agent setup. Running a CLI command triggers telemetry when a collector is configured.
Impact
Agent tooling is added without a separate user command, and configured telemetry can disclose the operator's email and command metadata to the collector.
Mechanism
The postinstall hook invokes skill installation and MCP registration. The CLI sends command metrics and logs, including operator email, to the configured OTLP endpoint unless telemetry is disabled.
Evidence
scripts/postinstall.mjsdist/cli.jsdist/skills/dr-cli/SKILL.md~/.claude/skills/<name>/~/.gemini/skills/<name>/~/.gemini/GEMINI.md~/.claude.json