How mujeebmohammed-gits/aigovernance-mcp works
This repository implements an AI governance and compliance control tower. An agent submits an AI-call context to the MCP orchestrator, which checks policy and agent compliance, dispatches an allowed call through a provider integration, and gathers the response. The orchestrator also coordinates audit, usage, and monitoring capabilities. A separate internal LLM API is reachable from the orchestrator. The tree advertises Jira, Confluence, and Slack integrations, but their wiring is not shown in sampled source; those remain separate optional branches. Evidence is limited to the provided tree and excerpts.
Main entry points: mcp_service.py MCP Orchestrator · integration_layer.py Provider Integrations · usage_tracker.py Usage Tracking · slack_integration.py Slack Integration
Components 18
Governance
- Policy Engine
policy_engine/policy_engine_main.py - Agent Registry
agent_registry/agent_registry.py - Provider Routing
provider_config/provider_config.py
Call Execution
- MCP Orchestrator
mcp_service/mcp_service.py - Provider Integrations
integration/integration_layer.py
Operations
- Audit Logging
audit/audit_logger.py - Usage Tracking
usage_tracker/usage_tracker.py - Usage JSONL
- Monitoring Dashboard
dashboard/dashboard.py - Configuration
config/config.py
Integrations
- OpenAI
- Anthropic
- Google Gemini
- Internal LLM API
internal_llm_api/server.py - Slack Integration
slack_integration.py - Jira Integration
jira_integration.py - Confluence Integration
confluence_integration.py
Other components
- Agent
Connections 15
- Agent → MCP Orchestrator: submits call
- MCP Orchestrator → Policy Engine: evaluates policy
- MCP Orchestrator → Agent Registry: checks compliance
- MCP Orchestrator → Provider Integrations: dispatches call
- Provider Integrations → OpenAI: calls provider
- Provider Integrations → Anthropic: calls provider
- Provider Integrations → Google Gemini: calls provider
- MCP Orchestrator → Internal LLM API: posts request
- MCP Orchestrator → Audit Logging: coordinates logging
- MCP Orchestrator → Usage Tracking: coordinates tracking
- Usage Tracking → Usage JSONL: persists events
- MCP Orchestrator → Monitoring Dashboard: coordinates monitoring
- Slack Integration → MCP Orchestrator: optional integration
- Jira Integration → MCP Orchestrator: optional integration
- Confluence Integration → MCP Orchestrator: optional integration