@telora/daemon@0.21.163
Agent orchestration daemon for Telora - spawns and manages Claude Code instances
AI Security Review
scanned 3d ago · by lpm-firewall-aiThe source shows a first-party agent configuration change that runs during daemon startup and can affect global Codex sessions, so a warning is warranted. The package has no install-time hook, and the inspected code does not establish credential theft or another concrete attack that supports blocking.
Decision evidence
public snapshot- The manifest has no install hook; its only publish lifecycle script builds and stamps the package. The CLI exposes harness setup through an explicit connect command.
- On daemon startup, provisioning refreshes the global Codex instruction file when products are configured and that file already exists; this refresh is not gated on a saved harness selection.
- The shared seed reaches interactive Codex and hosted daemon Codex agents, and adds Telora workflow instructions to their context.
- The writer appends or replaces the Telora-delimited block while retaining content around it. Initial harness setup is selected through connect, which defaults to Claude.
- Other inspected locations describe harness-specific settings, read-only audit guards, a resolution-agent completion handler, and a configured Lacuna install step; they do not establish another attack.
Source & flagged code
6 flagged · loading sourcePackage source references child process execution.
dist/update/self-update.jsView on unpkg · L21Package source invokes a package manager install command at runtime.
dist/update/self-update.jsView on unpkg · L12Findings
Affected versions and remediation
This report applies to @telora/daemon@0.21.163.
See all versions of @telora/daemon for other recorded verdicts.
- Review the evidence and your use of @telora/daemon@0.21.163 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.