instar@1.3.1380
Coherence infrastructure for self-evolving AI agents — on the Claude Code or Codex subscription you already have.
AI Security Review
scanned 4d ago · by lpm-firewall-aiThe migration automatically expands agent permissions and allows permission requests in project settings, creating a concrete agent capability risk. The behavior is tied to first-party post-update setup, so it warrants a warning rather than a block.
Decision evidence
public snapshot- On startup, a changed package version triggers the post-update migration.
- That migration edits an existing project-level .claude/settings.json file.
- The migration registers a PermissionRequest hook whose generated handler allows all permission requests.
- It also adds local tool permissions, including Bash, to the project allow list.
- The npm postinstall invokes a SQLite binding fixer; the permission changes occur in the runtime migration.
- The reviewed credential path uses Keychain storage and sends quota requests to api.anthropic.com.
- The settings migration skips when .claude/settings.json is absent; the source describes creating it during a later init.
- The credential caller uses the saved token for Anthropic quota requests, with no unrelated recipient shown in this flow.
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@1.3.1380.
See all versions of instar for other recorded verdicts.
- Review the evidence and your use of instar@1.3.1380 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.