Shell command execution detected (child_process).
Critical
- Code
- suspicious.dangerous_exec
- Location
- src/tools/context.ts:32
- Evidence
exec('git log --oneline -5', { cwd: dir, timeout: 3000 }, (err: any, stdout: string) => {
Security audit
Ask Me First
Security checks across malware telemetry and agentic risk
Overview
This avatar mostly matches its stated purpose, but it needs review because it can automatically speak for you, grants first-contact admin by default, requests broad Feishu permissions, and uses unsafe dynamic rule evaluation.
Before installing, preconfigure the real admin user, disable first-sender auto-admin unless you fully control first contact, grant Feishu least-privilege scopes, and avoid editable escalation rules until the `eval` implementation is replaced. Also review the workspace files because the avatar stores conversation-derived learning data and can automatically reply on your behalf.
VirusTotal
VirusTotal engine telemetry is currently stale for this artifact.
Static analysis
Detected: suspicious.dangerous_exec, suspicious.dynamic_code_execution
Shell command execution detected (child_process).
Critical
- Code
- suspicious.dangerous_exec
- Location
- src/tools/presence.ts:73
- Evidence
exec(`powershell -NoProfile -Command "${script.replace(/"/g, '`"')}"`, (err: any, stdout: string) => {
Dynamic code execution detected.
Critical
- Code
- suspicious.dynamic_code_execution
- Location
- src/escalation/rules.ts:63
- Evidence
const ok = eval(rule.condition.replace(/state\./g, 'safeState.').replace(/identity\./g, 'safeIdentity.'));