No concrete malicious chain, lifecycle abuse, or unrelated credential exfiltration is present. Its broad autonomous-agent and credential-proxy capabilities remain a meaningful user-runtime risk.
Static reason
One or more suspicious static signals were detected.
Trigger
User runs gcli/agent-cli, especially task-argument mode or interactive startup.
Impact
A compromised or prompt-injected model/provider could cause project changes or commands under the invoking user.
Mechanism
LLM-driven shell/file tools, local credential proxying, and background npm global update.
Evidence
package.jsondist/cli.jsREADME.md~/.claude/.credentials.json~/.codex/auth.json~/.agentos/config.json
Network endpoints4
console.anthropic.com/v1/oauth/tokenapi.anthropic.comauth.openai.com/oauth/token/backend-api/codex