instar@1.3.1357
Coherence infrastructure for self-evolving AI agents — on the Claude Code or Codex subscription you already have.
AI Security Review
scanned 5d ago · by lpm-firewall-aiSource inspection found automatic first-party hook updates during server migration, a guarded agent-extension lifecycle capability that warrants a warning. The inspected install and credential code did not establish malicious exfiltration or another blocking attack.
Decision evidence
public snapshot- On server startup, the package runs PostUpdateMigrator against the configured project.
- The migrator writes executable hook files into the Instar hook directory without a consent check at that migration point.
- The package includes commands and helpers that operate on Claude configuration and project hook locations.
- The postinstall hook runs a better-sqlite3 repair script; inspected code downloads its prebuild from the dependency's GitHub releases or rebuilds it.
- Inspected credential helpers read and write credentials through the operating system keychain; no unrelated recipient was established.
- The hook migration is first-party agent setup tied to the package's update flow; the inspected hook is described as blocking free-text credential prompts.
- The observed postinstall network request targets the better-sqlite3 project's release assets for native dependency repair.
- The inspected credential excerpts show local keychain access and storage, not credential transmission.
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@1.3.1357.
See all versions of instar for other recorded verdicts.
- Review the evidence and your use of instar@1.3.1357 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.