@telora/daemon@0.23.25
Agent orchestration daemon for Telora - spawns and manages Claude Code instances
AI Security Review
scanned 3d ago · by lpm-firewall-aiThe source shows automatic, continuing mutation of configured Claude agent settings to install a package-owned hook, which warrants a warning under the agent extension policy. The inspected hook invokes a local command, and the package manifest has no npm install lifecycle hook; the cited runtime-operation excerpts do not establish a separate attack.
Decision evidence
public snapshot- The manifest lists prepublishOnly as a build command; the inspected scripts show no install hook.
- Daemon startup refreshes Claude settings for configured products, and the refresher processes products with repository paths.
- The settings writer targets each repository’s .claude/settings.json and merges the package’s Claude settings when needed.
- The merged settings install a SessionStart hook that invokes the local telora-daemon session-state command.
- The flagged runtime excerpts show worktree installation, install-script analysis, configuration reload, and token parsing; they establish no separate attack behavior.
- The connector modules provide harness-specific agent and MCP configuration as part of the package’s setup flow.
- The SessionStart command runs a local package command; the inspected hook does not establish credential transfer or remote payload execution.
Source & flagged code
7 flagged · loading sourcePackage source references child process execution.
dist/update/self-update.jsView on unpkg · L27Package source invokes a package manager install command at runtime.
dist/update/self-update.jsView on unpkg · L17Findings
Affected versions and remediation
This report applies to @telora/daemon@0.23.25.
See all versions of @telora/daemon for other recorded verdicts.
- Review the evidence and your use of @telora/daemon@0.23.25 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.