instar@1.3.1406
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 supports a warning for automatic, version-triggered changes to project agent hooks that can affect later tool calls. The npm postinstall repair is separate, and the inspected telemetry and credential paths do not establish an installation beacon or credential exfiltration.
Decision evidence
public snapshot- On a version change, server startup runs the migrator, which adds Instar hooks to project Claude Code settings.
- The migrator registers an MCP pre-tool hook that runs an Instar script and can affect later tool calls.
- Baseline telemetry requires explicit CLI confirmation, defaults off, and checks feature consent before sending.
- The credential excerpts show OS keychain reads and writes; they do not show credentials being sent to a remote recipient.
- The other inspected integration locations concern project initialization, explicit setup or migration, a worktree marker, and atomic hook-file writing.
- The npm postinstall script runs a better-sqlite3 repair helper; the inspected hook configuration is performed later during server startup after a version change.
- No installation-time telemetry transfer was established; the inspected telemetry flow is opt-in and consent-gated.
Source & flagged code
17 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@1.3.1406.
See all versions of instar for other recorded verdicts.
- Review the evidence and your use of instar@1.3.1406 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.