MSKazemi/kubeintellect v2.0.2

on GitHub

KubeIntellect 2.0.2 rolls up agent hardening, the reflexion subsystem, the multi-version evaluation framework, and a complete V2 documentation overhaul since v2.0.1.

Agent & reflexion

• Strengthened coordinator prompt, token streaming, and kubectl safety
• Reflexion subsystem: cluster-identity resolution, mutation detection on outcomes, service-endpoint cross-check
• Orphan tool-call filler and expanded investigation playbooks

Evaluation framework

• Multi-version evaluation harness with 60+ scenarios across six categories and compare.py (paired t-tests + figures)
• Telemetry foundation (AgentTelemetry / AgentTelemetryEvent)
• --target / --categories CLI flags, per-scenario category tags, versioned EvalRecord

Observability

• Restructured logging and telemetry integration

Documentation

• New user guides: What you can ask, CLI Reference, API Reference, Troubleshooting, Operations, Glossary, Evaluation Harness
• Architecture & agent-behavior docs aligned with the code (4 tools incl. run_helm; 18 playbooks)
• Restructured site navigation (Use / Install / Deploy / Reference / Operations)

CI & packaging

• Deploy gated on the production environment + main branch
• Dependency and action-pin updates

Full changelog: v2.0.1...v2.0.2