Private · Production-ready

Orchestration, cost, and security for AI agents

AgentOps is the operations center: workflows and triggers, token and budget tracking, an agent marketplace, and enterprise security on the AegisAI core.

Self-host with Docker Compose · docs at /documentation/

Why it matters

Without an ops control plane, agents grow chaotically — in cost and in risk

🔗

No orchestration

Agent chains are wired by hand: no shared workflows, cron, or webhook triggers.

💸

No cost control

Tokens and latency are invisible per agent — budgets burn without alerts.

🧩

No reuse

Every team rebuilds agents from scratch — no marketplace or templates.

🕵️

Security blind spot

Shadow AI, prompt injection, and risky actions run without policies or HITL.

📊

No observability

Hard to see which agents are expensive, slow, or error-prone.

Built for platform and SecOps teams

From workflows to budgets and policies — one control plane for agents

🔗

Workflows

Visual and API agent chains, cron & webhooks, HITL pause and resume.

💰

Cost tracking

Tokens, latency, API calls, budgets, and alerts for expensive or slow agents.

🛒

Marketplace

Publish, rate, and install agents and templates for common jobs.

🤖

Registry & search

Registration, heartbeat, semantic search, and risk scoring from embeddings.

🛡️

Policy & Guard

OPA/Rego, Prompt Guard, and signed audit — security core from AegisAI.

⚠️

Anomalies

Hybrid ML + rules, Redis cache, runtime behavior control.

✋

Human-in-the-Loop

Approve risky workflow steps and agent actions.

📈

Observability

Cost and error dashboard, Prometheus / Grafana, alerts.

🔌

API & integrations

REST API, webhooks, SSO/OIDC, multi-tenant RBAC.

Product interface

Dashboard, Guard, HITL, and monitoring in one product

AgentOps Dashboard — KPI dashboard: agents, cost, policies, and HITL queue

Security built in

Multiple independent layers before an agent action executes

JWT / API key / SSO authentication

RBAC and tenant isolation

Prompt Guard middleware

OPA policy check + hybrid anomaly / risk score

HITL for critical operations and workflows

Signed audit + rate limits

Platform stack

FastAPI, OPA, Redis, PostgreSQL, Vite SPA, and observability built in

FastAPI REST + WebSocket API

Workflows Chains · cron · webhooks

Cost engine Tokens · budgets

OPA / Rego Policy decisions

PostgreSQL Agents, policies, audit

Redis Cache + Guard stats

Vite SPA Dashboard + i18n

Prometheus /metrics endpoint

25+ agent types

15+ policy templates

10 000 signed audit

3 multi-tenant

Quick start

Run locally in minutes with Docker Compose

# Run (private repo access required)
cd AgentOps
cp .env.example .env
docker compose up --build -d

# Health check
curl -s http://localhost:8001/api/v1/health

# Create a workflow (JWT + tenant API key)
curl -X POST http://localhost:8001/api/v1/workflows \
  -H "Authorization: Bearer $JWT" \
  -H "X-API-Key: $TENANT_API_KEY" \
  -d '{"name":"Daily Report","steps":[...]}'

API Dashboard