Skip to main content

Product Roadmap

Proxima Console follows a phased development model. Each phase delivers a working, tested feature set with full documentation and audit compliance.

Current Status​

Latest Release: v0.6.x (Operations & Fleet — incident response, fleet self-update, observability alerting)

Current Focus: Phase 6 — Polish & Production Readiness (v1.0.0)

Completed Phases​

Phase 1: Foundation (v0.1.0)​

  • Agent enrollment and inventory collection (host metadata, packages, containers, disks)
  • Multi-tenant backend with client/environment scoping
  • Console UI with host inventory dashboard
  • PostgreSQL data layer with migrations
  • NATS messaging for agent communication

Phase 2: Monitoring & Metrics (v0.2.0)​

  • System metrics collection (CPU, memory, disk, network, processes)
  • VictoriaMetrics integration with per-client data isolation
  • Time-series metric charts and dashboards
  • Integration collectors (PostgreSQL, Nginx, Docker)
  • Grafana dashboard provisioning

Phase 2.5: Auth, Process & Search (v0.2.5)​

  • JWT authentication with refresh tokens
  • Role-based access control (RBAC) with granular permissions
  • Google SSO (OIDC) integration
  • Audit logging for all API operations
  • PQL (Proxima Query Language) for host search
  • Process monitoring and listing

Phase 3: Change Detection & Logs (v0.3.0)​

  • File integrity monitoring (config files, system files)
  • Configuration drift detection
  • Log collection via NATS (journald, file-based)
  • VictoriaLogs integration for centralized log storage
  • NATS JWT credential system with per-host isolation
  • Vault integration for secrets management

Phase 4: Healthcheck & Compliance (v0.4.0)​

  • Healthcheck scanner with configurable rules
  • Compliance scoring framework (per-host and aggregate)
  • Credential management with Vault Transit encryption
  • Agent config sync (remote configuration push)
  • Webhook integrations (GitLab, GitHub, ArgoCD)

Phase 4.5: Alerting & Correlation (v0.4.1)​

  • Alertmanager webhook ingestion
  • Alert correlation with hosts and changes
  • Alert group management and silencing
  • Alert timeline views

Phase 5: MCP Server & AI Chat (v0.5.0)​

  • AI chat interface with 20+ infrastructure tools
  • LLM model routing (LiteLLM gateway)
  • MCP server for external LLM integration (Claude Desktop, Cursor, VS Code)
  • Conversation encryption (AES-256-GCM via Vault)
  • Runbook framework with 10 typed actions
  • Service desk integration (Jira ITSM, SLA tracking)
  • Terminal access: browser (WebSocket) and CLI (gRPC)
  • pc CLI tool: ssh, ls, scp, sessions, play, port forwarding
  • Session recording with asciicast v2 + Cloudflare R2
  • Multi-party sessions (observe/participate)
  • Kubernetes cluster inventory (13 resource types)
  • Database topology discovery (PostgreSQL primary/replica)
  • Passkey (WebAuthn) authentication
  • TOTP MFA

Phase 6: Operations & Fleet (v0.6.x)​

  • Daily briefing home page ("what needs attention") — prioritized list of items awaiting you, served by GET /api/v1/briefing
  • Per-client feature flags with launch defaults and admin toggle UI
  • Agent fleet management — version-skew visibility and canary/batched self-update rollouts (/api/v1/fleet/*) with pause/resume/abort and automatic pause-on-failure
  • Single POST /api/v1/agents/enroll enrollment with enroll-conflict (409) identity-takeover protection; per-agent NATS inbox isolation; reusable install tokens (max_uses)
  • L1 incident agent — native escalation engine, on-call schedules, deterministic triage context packs, AI-assisted RCA, a human-in-the-loop memory loop (prior-resolution retrieval), and a standalone Telegram ops bot
  • MCP server expansion — action tools (alert lifecycle, runbook executions, re-evaluation triggers, service desk, cross-entity PQL search), structured output with outputSchema, browseable resources, and invokable prompts (triage/diagnose/compliance/fleet workflows)
  • Agent distributed tracing — OTel spans exported over NATS-routed OTLP to a backend trace-ingest gateway (UI-controlled, off by default)
  • Agent self-log shipping — agents ship their own WARN+ logs to VictoriaLogs, trace-correlated
  • Observability alerting — Grafana Alerting fires a starter set of high-signal rules (API 5xx rate, p99 latency, NATS dead-letters, trace-ingest errors, dependency errors, DB pool saturation, backend-down, agent self-log errors) routed to Telegram
  • Agent disk buffer — durable, bounded, at-least-once outbound queue that spills to disk when NATS is unreachable and replays on reconnect
  • Kubernetes/ArgoCD GitOps production deployment (Helm charts for node-monitor and cluster agent)

In Progress​

Phase 6: Polish & Production Readiness (v1.0.0)​

  • Agent binary signing (GPG)
  • Data retention auto-purge policies
  • PII scrubbing for telemetry data
  • Per-client encryption key provisioning
  • PostHog product analytics integration
  • .deb and .rpm agent packages
  • Compliance documentation (SOC 2, ISO 27001, GDPR)
  • Production security hardening checklist

Infrastructure Topology & Network Discovery (v0.5.5)​

  • Network topology mapping (listening ports, established connections)
  • Service dependency graph (which hosts talk to which)
  • Infrastructure topology visualization
  • Network-aware alerting ("host A can't reach host B")

Future (Post v1.0)​

Kubernetes Deep Integration (v1.1)​

  • Kubernetes-specific metrics from kubelet and API server
  • Helm release tracking
  • ArgoCD integration for GitOps change detection
  • K8s RBAC audit
  • pc kube login — CLI access to clusters through Proxima proxy

AI-Powered Insights (v1.2)​

  • Anomaly detection on metrics
  • Correlation engine: changes linked to metric anomalies
  • Proactive recommendations ("CPU trending up 15% weekly, consider scaling")
  • Weekly infrastructure health reports
  • Infrastructure audit automation with LLM-powered analysis

Advanced Access (v1.3)​

  • Database access tunneling (pc db connect)
  • Remote port forwarding (pc ssh -R)
  • Access requests (just-in-time approval workflow)
  • Session recording search and analytics

Release Cadence​

  • Minor releases (v0.x.0): every 2-4 weeks during development
  • Patch releases (v0.x.y): as needed for bug fixes and security
  • Stable release (v1.0.0): targeting production readiness with full security audit

All releases follow semantic versioning and are documented in the changelog.