instar@1.3.1405
Coherence infrastructure for self-evolving AI agents — on the Claude Code or Codex subscription you already have.
AI Security Review
scanned 2d ago · by lpm-firewall-aiSource inspection confirms that successful updates can trigger first-party hook installation and default-on Codex auto-arming, which can activate package-owned hooks without the usual human trust action. This is a guarded extension lifecycle risk; the inspected source does not support a publish block.
Decision evidence
public snapshot- The manifest declares a postinstall command that runs a SQLite repair script.
- After a successful update, the updater invokes the new CLI's migration command.
- The migration flow installs Instar-owned hooks and updates agent settings.
- Codex hook auto-arming defaults to on and uses a tmux trust driver when Codex is detected; a configuration setting can disable it. The source says the hooks otherwise await a human trust action.
- The inspected setup locations cover project documentation, a command-driven hook update, Playwright MCP setup, worktree metadata, relay script writing, and Codex hook configuration.
- The cited credential excerpts show local OS keychain reads and writes; they do not establish an external recipient.
- The inspected postinstall declaration does not invoke the post-update migrator.
- The reviewed credential and quota excerpts do not establish credential exfiltration or an unrelated account action.
Source & flagged code
18 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgPackage source references child process execution.
dist/threadline/PipeSessionSpawner.jsView on unpkg · L15Package source references shell execution.
Findings
Affected versions and remediation
This report applies to instar@1.3.1405.
See all versions of instar for other recorded verdicts.
- Review the evidence and your use of instar@1.3.1405 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.