Recent Releases
AI Prediction Tracking & Accuracy Ledger
Establishes a historical prediction ledger allowing future outcomes to be evaluated against forecasts, risks, and recommendations generated by the Xeyria Platform.
- ›Exposes prediction types: forecast, risk, and recommendation with full lifecycle states (pending, confirmed, rejected, expired).
- ›Integrates automated lookup evaluations directly within backend workers to audit forecast accuracy.
- ›Enforces database-level immutability on prediction data after creation via customized PostgreSQL triggers.
- ›Enables full Row-Level Security (RLS) on prediction records to guarantee secure tenant isolation.
Project Risks Telemetry & Scan Engine
Introduces a background scanning service that automatically detects project health declines, inactive files, pending bottlenecks, and answered question delays.
- ›Stale Questions Scans: Flags risks when project questions remain open for more than 14 days.
- ›Inactivity Monitoring: Warns developers if no file modifications or events occur in 7 days.
- ›Knowledge Stagnation: Identifies workspaces with no document updates for over 30 days.
- ›Severity Escalation: Automatically escalates open bottlenecks (Day 8 Medium -> Day 16 High -> Day 31 Critical).
- ›Interactive Risks Panel: A premium sidebar to browse, snooze, or dismiss active risks with optimistically updated UI states.
Subscription Billing & Razorpay Integration
Deploys our server-side aggregated Billing Portal integrated with Razorpay for secure checkout, invoices, and payment lifecycle tracking.
- ›Secure aggregated backend GET /api/billing routes to avoid exposing database schema and bypass client-side RLS limits safely.
- ›Support for Starter and Pro plans, trial checking, and automated payment retries.
- ›Timeline events tracking billing history (orders, payments, invoices, plan changes).
- ›Disabled broken invoice URL clicks, replacing them with professional "Coming Soon" tooltips.
Core Memory Engine & Context Sync
The foundation of the Xeyria Project Intelligence platform. Introduces the local indexers and memory graphs that maintain context.
- ›Project Memory: Permanent log storing architectural principles, developer rules, and workspace objectives.
- ›Context Inspector: A panel detailing loaded workspace requirements and active developer code ranges.
- ›Workspace Sync: Automatic vector synchronization of developer edits and configuration changes.
- ›Theme System: Fully custom dark-mode-only dashboard with oklch-based color tokens.
Upcoming Roadmap
We are committed to building long-term continuity for builders. Here is what our engineering team is actively working on:
Multi-Agent Orchestration Layer
Enable multiple specialized AI agents (e.g. Architect, Tester, Writer) to communicate, collaborate, and share the same synchronized workspace context without crossing lines.
Local Embedding Model Support
Integrate local vectorization layers running on-device. This allows workspace indexing to complete without transmitting any metadata chunks outside your machine, ensuring absolute code privacy.