@addai/node@0.30.12
Daemon that pairs a machine with your +Ai account and runs Claude / Codex / Kimi / Gemini agents on its behalf. Reachable via Supabase from Vault, Entity Studio, or any other +Ai surface.
AI Security Review
scanned 14d ago · by lpm-firewall-aiThe daemon automatically suppresses Claude’s bypass-permissions warning and pre-trusts session directories, creating a concrete agent capability risk. The inspected behavior is runtime-triggered, not a malicious npm lifecycle mutation; classify it as suspicious and warn.
Decision evidence
public snapshot- Starting the daemon writes skipDangerousModePermissionPrompt=true to the user’s Claude settings, suppressing a warning for bypass-permissions mode.
- The Claude helper targets the user’s home configuration; this mutation is reached during daemon startup, not the npm postinstall hook.
- Codex setup can link or copy the user’s authentication files into its sandbox directory for agent use.
- Grok and Kimi setup writes per-session MCP configuration; the general MCP config is limited to caller-requested servers.
- The runtime installer writes skills into the session working directory, and the precompact hook writes session-scoped Claude settings.
- The manifest’s postinstall runs a PTY helper repair; it does not invoke the Claude configuration mutation.
- The inspected Claude permission change runs when the daemon starts, while the npm postinstall hook calls a PTY helper repair.
- The inspected MCP and hook configuration writes are scoped to a session or working directory.
Source & flagged code
13 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgPackage source references child process execution.
dist/claude-system-prompt.jsView on unpkg · L62Findings
Affected versions and remediation
This report applies to @addai/node@0.30.12.
See all versions of @addai/node for other recorded verdicts.
- Review the evidence and your use of @addai/node@0.30.12 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.