@tekdi/coding-analytics
Monitor Cursor & Claude Code usage and send analytics to PostHog. Uses hooks to capture events.
This report shows the evidence for @tekdi/coding-analytics@1.0.6. 0 of 2 known versions are blocked or flagged. See all versions.
AI Security Review
scanned 6d ago · by lpm-firewall-aiThe analytics purpose and explicit hook installation are legitimate, but the shipped defaults enable response-text capture while the README says responses are not captured by default. This disclosure mismatch makes the runtime telemetry suspicious.
Decision evidence
public snapshot- The package describes itself as capturing Cursor and Claude Code events for PostHog analytics.
- Hook and cron setup runs through the explicit `coding-analytics install-hooks` command.
- Shipped config defaults enable prompt, response, and agent-thinking capture.
- Hook handlers include configured response text and tool inputs in queued events.
- The README sample disables response capture and says AI response content is not captured by default, contrary to the shipped default.
- Events are sent using the configured PostHog API key and host.
- Analytics collection and PostHog transmission match the package’s stated purpose.
- The hooks and scheduled flush are installed only when the user runs the install command.
- The PostHog destination is configurable, and file-content capture defaults to false.
- The README describes secret redaction and hashed or opaque user IDs.
Source & flagged code
6 flaggedPackage source references weak cryptographic algorithms.
dist/event-id.jsView on unpkg · L56Findings
Affected versions and remediation
This report applies to @tekdi/coding-analytics@1.0.6.
The versions list below shows the verdict for each recorded version.
- Review the evidence and your use of @tekdi/coding-analytics@1.0.6 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.