How yellowscorpiondpiii/capstone-mira works
Mira is a multi-agent technical program management platform. A client submits a message to the application, which dispatches work through the orchestrator to planning, risk, reporting, governance, roadmap, and tool-recommendation agents; workflows can combine these stages and return results. Agents use a shared LLM interface with routed cloud or local backends. The product also advertises six collaboration adapters and webhook/event processing. Source evidence is partial, so adapter wiring and several documented subsystems are shown without inferred call chains.
Main entry points: orchestrator_agent.py Workflow orchestrator · base_agent.py Agent runtime · app.py Mira application · router.py LLM router
Components 29
Application flow
- Mira application
mira/app.py - Configuration
mira/config/settings.py
Workflow agents
- Workflow orchestrator
mira/agents/orchestrator_agent.py - Project planning
mira/agents/project_plan_agent.py - Risk assessment
mira/agents/risk_assessment_agent.py - Status reporting
mira/agents/status_reporter_agent.py - Governance review
mira/agents/governance_agent.py - Roadmapping
mira/agents/roadmapping_agent.py - Tool recommendation
mira/agents/tool_recommender_agent.py - Scaling dashboard
- Talent orchestration
mira/agents/talent_orchestrator.py - Agent runtime
mira/core/base_agent.py
LLM services
- Output schemas
mira/llm/schemas.py - LLM router
mira/llm/router.py - LLM interface
mira/llm/base.py - Anthropic backend
mira/llm/anthropic_llm.py - OpenAI backend
mira/llm/openai_llm.py - Ollama backend
mira/llm/ollama_llm.py
Integrations
- Service adapters
mira/integrations - Adapter contract
mira/integrations/base_integration.py
Events and governance
- Message broker
mira/core/message_broker.py - Webhook handler
mira/core/webhook_handler.py - Human approval handling
governance/hitl_handler.py - Governance risk assessor
governance/risk_assessor.py - Logging
mira/utils/logging.py - Structured logging
mira/utils/structured_logging.py - Graceful shutdown
mira/utils/shutdown_handler.py
Other components
- Workflow client
- Collaboration services
Connections 34
- Workflow client → Mira application: submits work
README.md - Mira application → Workflow orchestrator: registers and dispatches
mira/app.py - Workflow orchestrator → Project planning: routes plan tasks
mira/agents/orchestrator_agent.py - Workflow orchestrator → Risk assessment: routes risk tasks
mira/agents/orchestrator_agent.py - Workflow orchestrator → Status reporting: routes report tasks
mira/agents/orchestrator_agent.py - Workflow orchestrator → Governance review: routes governance checks
mira/agents/orchestrator_agent.py - Workflow orchestrator → Roadmapping: routes roadmap tasks
mira/app.py - Workflow orchestrator → Tool recommendation: routes recommendations
mira/app.py - Workflow orchestrator → Message broker: publishes approvals
mira/agents/orchestrator_agent.py - Mira application → Message broker: starts broker
mira/app.py - Mira application → Webhook handler: starts when enabled
mira/app.py - Mira application → Configuration: reads configuration
mira/app.py - Mira application → LLM router: configures model routing
mira/app.py - Project planning → Agent runtime: uses runtime
mira/agents/project_plan_agent.py - Risk assessment → Agent runtime: uses runtime
mira/agents/risk_assessment_agent.py - Status reporting → Agent runtime: uses runtime
mira/agents/status_reporter_agent.py - Governance review → Agent runtime: uses runtime
mira/agents/governance_agent.py - Roadmapping → Agent runtime: uses runtime
mira/agents/roadmapping_agent.py - Tool recommendation → Agent runtime: uses runtime
mira/agents/tool_recommender_agent.py - Agent runtime → LLM router: generates responses
- LLM router → Anthropic backend: routes requests
mira/llm/router.py - LLM router → OpenAI backend: routes requests
- LLM router → Ollama backend: routes local requests
mira/llm/router.py - Anthropic backend → LLM interface: implements interface
mira/llm/anthropic_llm.py - OpenAI backend → LLM interface: implements interface
mira/llm/openai_llm.py - Ollama backend → LLM interface: implements interface
mira/llm/ollama_llm.py - Agent runtime → Output schemas: accepts output schema
- Service adapters → Adapter contract: implements adapter contract
mira/integrations/base_integration.py - Service adapters → Collaboration services: syncs collaboration data
README.md - Human approval handling → Governance risk assessor: assesses approval risk
governance/hitl_handler.py - Human approval handling → Message broker: notifies approvals
- Mira application → Logging: configures logging
mira/app.py - Structured logging → Logging: builds on logging
mira/utils/structured_logging.py - Graceful shutdown → Logging: logs shutdown
mira/utils/shutdown_handler.py