Tutorials
Practical, end-to-end guides based on real smoke tests from the Kavach codebase. Each tutorial demonstrates actual API calls and MCP tool usage.
Local Developer Workflow
Start Keycloak, REST, MCPO, Studio, Neo4j, demo seed data, and worker health in the supported local workflow.
Prerequisites
Python 3.12+ · uv · Docker for infrastructure
Docker Deployment
Stand up the full stack: REST API, Kavach Studio, and Neo4j with managed datastores.
Prerequisites
Docker installed
REST Control Plane
Explore the full REST API surface: health checks, evaluations, governance comparisons, experiments, and job management.
Prerequisites
Docker Compose running · API on localhost:8000
Review Governance Decisions
Inspect a decision outcome, validate its evidence graph, follow ontology lineage, and record a defensible review conclusion in Kavach Studio.
Prerequisites
Kavach Studio running · Seeded data or a persisted decision
Kavach Studio - Experiment Management
Register governed candidates, run evaluations, compare configurations and metrics, and inspect leaderboard recommendations in Studio.
Prerequisites
Kavach Studio running · Seeded data or governed registry assets
Kavach Studio - Replay Management
Reproduce a historical execution from frozen evidence, follow worker execution, and review comparison and drift results.
Prerequisites
Kavach API and worker running · Replayable historical execution
MCP Server Basics
Learn the Kavach MCP foundations, from direct calls and stdio to native Streamable HTTP and MCPO.
Prerequisites
Kavach API running · Python 3.12+
Native MCP Streamable HTTP
Connect MCP-aware clients over HTTP with a Keycloak bearer token, then validate the setup in MCP Inspector.
Prerequisites
Kavach local stack running · Python 3.12+
MCP Controlled Writes
Submit evaluations, create experiments, and manage jobs through the MCP server with full audit trails.
Prerequisites
MCP server basics · Kavach API running
Governance Insights
Explore experiment insights, candidate explanations, drift analysis, investigations, and report generation.
Prerequisites
Experiments and evaluations created · API running