@markus-global/cli@0.9.6
Markus — AI Digital Workforce Platform
AI Security Review
scanned 3mo ago · by lpm-firewall-aiNo concrete malicious or unconsented install-time behavior was found. The package nevertheless provides explicit-user agent configuration and browser-automation capability, warranting a warn verdict under the stated policy.
Decision evidence
public snapshot- Explicit `markus install <platform>` installs and initializes external agent CLIs.
- It writes Markus URL/token into OpenClaw or Hermes configs and copies an OpenClaw skill.
- Bundled browser extension has debugger and `<all_urls>` permissions, but is not install-time executed.
- `package.json` has no preinstall/install/postinstall hooks.
- Agent-platform changes occur only after an explicit CLI command.
- API client targets configured/local Markus server; no credential-exfiltration path found.
- Update downloads are user-invoked `markus update` behavior.
Source & flagged code
11 flaggedPackage contains a possible secret pattern.
dist/web-ui/assets/index-BMzsi3-i.jsView on unpkg · L10Hardcoded password in dist/web-ui/assets/index-BMzsi3-i.js
dist/web-ui/assets/index-BMzsi3-i.jsView on unpkg · L19Findings
Affected versions and remediation
This report applies to @markus-global/cli@0.9.6.
See all versions of @markus-global/cli for other recorded verdicts.
- Review the evidence and your use of @markus-global/cli@0.9.6 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.