Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.medera.info/llms.txt

Use this file to discover all available pages before exploring further.

​
2026-03-02 — RLS + audit hardening

  • Migration 158 — FORCE ROW LEVEL SECURITY completeness sweep
  • Migration 157 — GIN indexes (jsonb_path_ops) on JSONB columns
  • Migration 156 — JSONB defaultTo() audit (prevents double-stringified defaults)
  • Migration 155 — Dynamic FK index detection
  • Migration 154 — Standardized RLS on 12 tables — 4-policy has_tenant_access()

​
2026-03-02 — Workflow engine + deployments

  • 19-file workflow engine totaling 7,180 lines
  • WorkflowExecutor (471 L), ExecutionContext (354 L), WorkflowQueue (285 L) — Bull on Redis DB 2, 5 concurrent workers, 3-attempt retry with 5 s exponential backoff
  • Graceful sync fallback if Redis unavailable
  • Real-time Socket.IO events: workflow:execution:started, workflow:node:executing, workflow:node:completed, workflow:ehr-sync, workflow:execution:paused/completed/failed

​
2026-03-02 — Sandbox-first deployment platform

  • Sandbox URLs: https://sandbox.medera.info/{org}/{agent}-v{ver} with 72-hour auto_expire_at
  • Production URLs: https://agents.medera.info/{org}/{agent}
  • Migration 138 — agent_deployments table with 12 statuses
  • Migration 140 — deployment_audit_log (WORM) — migration 145 fixed FK from RESTRICT → SET NULL for WORM compat
  • Migration 144 — deployment_preview_access (token-based sandbox sharing)
  • Migration 148 — deployment_metrics (time-bucketed)
  • Migration 149 — deployment_alerts (threshold-based)
  • Migration 152 — Auto-rollback config + alert resolution

​
2026-02-10 — Security fixes

  • Hardcoded service keys removed (4 files, 17 occurrences)
  • Hardcoded DB credentials removed from templates.js and notifications.js
  • Auth rate limiting added (10 req / 15 min)
  • PII console.log statements purged
  • Patient PHI fields (SSN, insurance, address) encrypted AES-256-GCM
  • Phone + email hashed for search

​
2026-03-02 — Mock AI safety

  • 6 methods that returned fabricated clinical data replaced with unavailable status (confidence: 0.0, requires_human_review: True)
  • Crisis detection: all crisis_detected=False changed to None; escalation_required defaults to True
  • 11 bare except Exception clauses fixed with proper logging across 9 files