@clawos-dev/clawd@0.2.592
Standalone clawd daemon — Claude Code (and future Codex) session server over WebSocket
AI Security Review
scanned 27d ago · by lpm-firewall-aiThis is a user-started Claude session daemon that rewrites its own persona agent files and forwards local Anthropic settings into a guest spawn, which is guarded first-party agent setup rather than an install hook. No hidden exfiltration or unconsented foreign control-surface write was found, so the package should be warned and not blocked.
Decision evidence
public snapshot- package.json exposes only the clawd bin at dist/cli.cjs and has no preinstall, install, or postinstall script.
- The daemon replaces persona paths CLAUDE.md, .mcp.json, .claude/skills, .claude/agents, and mcp when it refreshes a persona.
- readOwnerCredentials loads ~/.claude/settings.json and the daemon injects matching owner credentials into the local guest spawn.
- The pi MCP bridge calls pi.registerTool when that extension is loaded.
- The flagged mcp-server hits are schema comments naming CLAUDE.md and .claude/skills, not install-time writes.
- No registry lifecycle hook runs this code during npm install.
- Copied Anthropic settings stay in the local guest spawn rather than being posted to a package-controlled host.
- publish.sh deploys an extension only when a user passes an extension directory, persona root, and env.
Source & flagged code
19 flagged · loading sourcePackage source references child process execution.
dist/app-builder-plugin/mcp-server.cjsView on unpkg · L21117Source file is highly similar to a previously finalized malicious package; route for source-aware review.
dist/app-builder-plugin/mcp-server.cjsView on unpkgPackage contains a possible secret pattern.
dist/app-builder-plugin/mcp-server.cjsView on unpkg · L24996Findings
Affected versions and remediation
This report applies to @clawos-dev/clawd@0.2.592.
See all versions of @clawos-dev/clawd for other recorded verdicts.
- Review the evidence and your use of @clawos-dev/clawd@0.2.592 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.