@clawos-dev/clawd@0.2.649
Standalone clawd daemon — Claude Code (and future Codex) session server over WebSocket
AI Security Review
scanned 10d ago · by lpm-firewall-aiSource inspection found no confirmed malicious behavior. The flagged agent-related writes are bounded to session metadata or a caller-named project operation, and the observed logging endpoint is part of the ClawOS product flow.
Decision evidence
public snapshot- package.json exposes a CLI and its scripts end without an install lifecycle hook.
- The CLI appends turn-duration metadata only to an existing Claude session file; its runner calls the helper when a turn ends.
- The three app-builder bundles resolve a project collection script, and the publish handler is registered as an MCP tool using configured project context.
- The flagged dynamic code creates a generated validator and provides a setImmediate compatibility helper.
- The guest viewer sends a supplied bearer token while omitting ambient cookies; the bridge registers configured MCP tools and dispatch listing uses the daemon URL and token.
- Daemon logging targets the ClawOS product log endpoint, includes owner and device identifiers, and has an explicit off setting.
Source & flagged code
15 flagged · loading sourcePackage source references child process execution.
dist/app-builder-plugin/mcp-server.cjsView on unpkg · L30858Package source references dynamic code evaluation.
dist/app-builder-plugin/mcp-server.cjsView on unpkg · L9420Source file is highly similar to a previously finalized malicious package; route for source-aware review.
dist/app-builder-plugin/mcp-server.cjsFindings
Affected versions and remediation
This report applies to @clawos-dev/clawd@0.2.649.
See all versions of @clawos-dev/clawd for other recorded verdicts.
- Review the evidence and your use of @clawos-dev/clawd@0.2.649 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.