Monitor and manage your AWS Bedrock agents from a single dashboard.
BedRex provides real-time visibility into your Amazon Bedrock agent fleet — inventory, topology, CloudWatch metrics, cost analysis, and configuration tracking across multiple accounts and regions.
Browse all your Bedrock agents across regions with filtering by status, model, region, and type. Each card shows action groups, knowledge bases, guardrail status, and 24h invocation counts.
Switch to a compact table layout with sortable columns. Click any row to open the detail drawer with full agent configuration, invoke capabilities, and snapshot history.
Inspect individual agents — view action groups, knowledge bases, guardrail config, foundation model, and version history. Invoke agents directly with custom prompts and compare configuration snapshots.
Interactive force-directed graph showing relationships between agents, action groups, knowledge bases, guardrails, and foundation models. Filter by node type, search nodes, and switch between layout directions.
KPI summary bar with total/active agents, invocations, errors, and estimated daily cost. Includes model distribution chart, status breakdown, pricing tier comparison, and per-model cost projections.
Cost estimation table with projectable token volumes (1K–1M tokens/day), per-agent CloudWatch metrics with latency, error rates, and real cost data. Expandable rows with time-series sparklines.
- Agent Inventory — View all Bedrock agents across accounts and regions with filtering, search, and sorting
- Topology Map — Interactive graph visualization of agents, action groups, knowledge bases, and guardrails
- CloudWatch Metrics — Real-time invocation counts, latency, error rates, and token usage from AWS CloudWatch
- Cost Analysis — Per-model and per-agent cost estimation based on actual token usage and Bedrock pricing
- Configuration Snapshots — Track agent configuration changes over time with diff views
- Multi-Account Support — Monitor agents across multiple AWS accounts using cross-account IAM roles
- Agent Invocation — Test agents directly from the dashboard with custom prompts
git clone https://github.com/bestcloudforme/bedrex.git
cd bedrex
cp .env.example .env
# Edit .env with your AWS_ACCOUNT_ID and AWS_REGION
docker compose upDemo mode: Set
USE_MOCK_DATA=truein.envto explore the UI without AWS credentials.
Prerequisites: Node.js 22+, npm 10+
git clone https://github.com/bestcloudforme/bedrex.git
cd bedrex
npm install
npm run devFrontend: http://localhost:5173 | Backend: http://localhost:3001
| Variable | Default | Description |
|---|---|---|
AWS_ACCOUNT_ID |
— | Your 12-digit AWS account ID (required for live mode) |
AWS_REGION |
us-east-1 |
Default AWS region for Bedrock API calls |
USE_MOCK_DATA |
false |
Enable demo mode without AWS credentials |
PORT |
3000 |
Port for the web UI (Docker) |
LOG_LEVEL |
info |
Log verbosity: trace, debug, info, warn, error |
CORS_ORIGIN |
http://localhost:5173 |
Allowed origins (comma-separated) |
CACHE_ENABLED |
true |
Toggle in-memory caching |
BedRex uses the standard AWS credential chain. Options:
- AWS CLI profiles — Configure with
aws configure(recommended for local dev) - Environment variables — Set
AWS_ACCESS_KEY_IDandAWS_SECRET_ACCESS_KEY - Docker volume mount — docker-compose mounts
~/.awsautomatically
Required IAM permissions:
bedrock:ListAgents,bedrock:GetAgentbedrock:ListAgentActionGroups,bedrock:ListAgentKnowledgeBasesbedrock-agent-runtime:InvokeAgentcloudwatch:GetMetricDatasts:GetCallerIdentity
bedrex/
├── packages/
│ ├── backend/ Express + TypeScript + AWS SDK v3
│ ├── frontend/ React 19 + Vite + TailwindCSS v4
│ └── shared/ Shared types and constants
├── infra/ AWS CDK (optional cloud deployment)
├── docker-compose.yml
└── .env.example
Tech Stack:
- Frontend: React 19, TypeScript, TailwindCSS v4, Recharts, React Flow, TanStack Query, Zustand
- Backend: Express, TypeScript, AWS SDK v3 (Bedrock, CloudWatch, STS)
- Build: Turborepo, Vite, tsx
- Infrastructure: Docker Compose, nginx (optional: AWS CDK)
- Fork the repository
- Create a feature branch (
git checkout -b feature/my-feature) - Make your changes
- Run
npm run buildto verify TypeScript compilation - Submit a pull request
MIT — see LICENSE for details.





