v2026.03.29

This release adds AI assistant integration (MCP), PostHog analytics, error tracking with GitHub Issue sync, infrastructure sidecars, and direct Odoo management from the dashboard. Built across phases 51-83 between 2026-03-07 and 2026-03-29.

MCP server (Phases 64-69, 83)

Model Context Protocol server for connecting AI assistants to manage Odoo instances.

• OAuth 2.1 authorization server with PKCE, device flow, and dynamic client registration
• 100+ tools across 14 categories (ORM, search, metadata, SQL, shell, files, modules, system, context, monitoring, export, attachments, reports, platform)
• Per-API-key permissions: read-only mode, tool allowlists, category scoping, instance scoping
• Knowledge Layer: notes, bookmarks, saved filters with auto-context injection at session start
• Field metadata caching with singleflight dedup
• Smart field selection scoring for query optimization
• Redis rate limiter with circuit breaker and per-category limits
• Daily usage tracking with buffered writes
• Invite links for onboarding external users
• SSE and Streamable HTTP transports, plus stdio proxy binary for local clients

PostHog analytics (Phases 81-82)

Per-instance analytics dashboards powered by PostHog.

• Revenue: KPIs, trend charts, top customers, currency breakdown
• Active users (DAU/WAU/MAU) with session replay deep links
• Web vitals: LCP, INP, CLS with ratings
• Slow RPC tracking and performance trends
• Error tracking via PostHog events
• Audit trail with severity classification and replay links
• Productivity metrics with per-user scoring
• Module usage analytics
• Background cache warming every 15 minutes
• Config reads from org_settings DB, no env vars needed

Odoo management (Phases 51-56)

Direct Odoo administration from the dashboard over SSH.

• User management: create, update, deactivate, group assignment, impersonation
• Module management: install, upgrade, uninstall with dependency visualization and operation history
• Scheduled actions: list, toggle, run on demand
• System parameters: read and write ir.config_parameter values
• Odoo logs: search, filter, tail
• ocore_connect module: token-based auth, auto-deploys on first use
• Per-instance semaphore: max 3 concurrent RPC calls

Infrastructure sidecars (Phase 57)

Optional services deployed alongside each Odoo instance via docker compose.

• PgBouncer: connection pooling with scram-sha-256 auth and wildcard DB support
• Redis: session storage for multi-worker deployments
• S3/MinIO: filestore offloading with automatic migration
• Toggle on/off from the dashboard, auto-enable based on resource thresholds

Error tracking (Phases 58-60)

Automatic error detection from Odoo logs.

• Python traceback parsing and SHA-256 fingerprint deduplication
• Hourly occurrence histograms
• GitHub Issue sync: auto-create on new error, close on resolution, reopen on regression
• Cooldown and concurrency handling for high-volume error streams
• Frontend: filterable list, area charts, detail panel with stack trace

GitHub App (Phase 63)

• Multiple installations per org (different GitHub accounts/orgs)
• Automatic API routing based on repository owner

Security and fixes

• Open redirect prevention in auth flows
• CORS hardening
• Token rotation with replay detection
• Per-org session policies
• Dependency patches (fast-xml-parser, dompurify, jsonparser, Next.js)
• Per-user SSE connection limit (max 10) with atomic counting
• DELETE operation rate limiting (10/min per user)
• Sensitive field markers on 80+ credential fields in DB schemas
• DB indexes on frequently-queried status/timestamp fields