v0.3.0

Root-cause evidence release.

Added

• Root cause decisions now include deterministic confidence levels: high, medium, or low.
• Root cause decisions now include evidence strings assembled from health status, latency, target, criticality, messages, and dependency relationships.
• FULL MODE now prints confidence and evidence under the rule engine section.
• PANIC MODE keeps the strict four-line output while adding confidence inline on the ROOT CAUSE line.
• Roadmap now includes signal adapters for Datadog, CloudWatch, git history, and DB state.

Notes

• This keeps the engine rule-based and lightweight while making root-cause output more trustworthy.
• External signal ingestion is intentionally roadmap-only for now; the implemented change focuses on evidence from existing health/dependency data.

v0.2.1

Documentation cleanup release.

v0.2.0

Dual-mode incident response release.

Added

• FULL MODE via panic check --full for engineering/debug visibility.
• PANIC MODE via panic emergency for strict four-line outage decisions.
• Deterministic single-cause root cause decision model with rule names and explanations.
• Mandatory prioritized NEXT ACTION output for emergency decisions.
• Impact line generation for panic-mode output.
• Normalized full-mode statuses using OK, DEGRADED, and DOWN.
• Dependency status section in full-mode output.
• Optional logSummary field in service config for full-mode log context.
• GET /emergency backend endpoint for panic-mode decision data.
• emergency npm script for local development.

Changed

• Bumped package version to 0.2.0.
• Updated README around dual-mode product architecture and npm/Bun usage examples.
• Enriched example and demo configs with dependency and log summary metadata.

Notes

• FULL MODE is intentionally detailed and calm.
• PANIC MODE is intentionally minimal and must not include extra diagnostic noise.
• Rule-based decisions remain deterministic; no AI or ML inference is used.

v0.1.0

Initial MVP release of Panic Tool.

Added

• CLI package published as @rekl0w/panic-tool with panic binary.
• panic check command for terminal-friendly multi-service health reports.
• panic status command for compact incident status summaries.
• panic incident command for human-readable triage output.
• panic init command to create panic.config.json from the example config.
• HTTP health checks with expected status support.
• TCP health checks for services such as databases, Redis, queues, brokers, and internal dependencies.
• Configurable timeout and latency warning thresholds.
• Dependency-aware service config via dependsOn.
• Rule-based probable root cause heuristics, including DB/API, Redis/queue, and high-latency patterns.
• Actionable recovery suggestions focused on what to do during an incident.
• Lightweight Hono backend with JSON endpoints:
  • GET /
  • GET /health
  • GET /status
  • GET /incident
• Demo config using public endpoints plus an intentionally failing TCP target for quick testing.
• TypeScript declarations for package consumers.
• GitHub Actions CI workflow for typechecking.
• Open-source project docs, contribution guide, MIT license, and npm package metadata.

Notes

• This is intentionally not an observability platform, dashboard, tracing system, log collector, alerting system, or Sentry clone.
• The MVP focuses on fast production incident triage: what is broken, what is probably causing it, and what to do now.
• Bun 1.3.11+ is required at runtime because the CLI binary targets Bun.