How mosami00/agentic-test-system works
Agentic Test Generation is a Python system for generating and maintaining unit tests. Its main workflow accepts a repository run through the CLI, analyzes Git changes and source symbols, plans targets, generates candidates, validates them, runs sandboxed tests, diagnoses failures, and reports results. Workflow state and events are persisted in SQLite. The graph shows the documented coordinator and major domain stages; some subsystem implementations were not excerpted, so their internal links are omitted or marked unverified.
Main entry points: service_factory.py Service Factory · nodes.py Workflow Nodes · engine.py Workflow Engine · service.py Repository Analysis
Components 25
User Entry
- CLI App
src/agentic_test/cli/app.py - Service Factory
src/agentic_test/cli/service_factory.py
Workflow Control
- Workflow Engine
src/agentic_test/workflow/engine.py - Workflow Nodes
src/agentic_test/workflow/nodes.py - Workflow Recovery
src/agentic_test/workflow/recovery.py - Workflow State
src/agentic_test/core/state.py
Analysis Planning
- Repository Analysis
src/agentic_test/analysis/service.py - Git Inspection
src/agentic_test/analysis/git_service.py - AST Analyzer
src/agentic_test/analysis/ast_analyzer.py - Test Discovery
src/agentic_test/analysis/discovery.py - Execution Planner
src/agentic_test/planning/planner.py - Planning Rules
src/agentic_test/planning/rules.py
Test Lifecycle
- Test Generation
src/agentic_test/generation/service.py - Prompt Context
src/agentic_test/generation/context.py - LLM Adapter
src/agentic_test/generation/litellm_service.py - Candidate Validation
src/agentic_test/validation/pipeline.py - Sandbox Execution
src/agentic_test/execution/service.py - Coverage Metrics
src/agentic_test/execution/coverage.py - Failure Diagnosis
src/agentic_test/diagnosis/service.py - Triage Rules
src/agentic_test/diagnosis/rules.py - LLM Triage
src/agentic_test/diagnosis/llm_classifier.py
State Persistence
- SQLite Database
src/agentic_test/storage/database.py - Run Repository
src/agentic_test/storage/repository.py - Event Store
src/agentic_test/storage/events.py
Other components
- Engineer
Connections 37
- Engineer → CLI App: invokes
- CLI App → Service Factory: configures services
src/agentic_test/cli/app.py - CLI App → Workflow Recovery: recovers runs
src/agentic_test/cli/app.py - CLI App → Git Inspection: inspects repository
src/agentic_test/cli/app.py - Service Factory → Repository Analysis: constructs
src/agentic_test/cli/service_factory.py - Service Factory → Execution Planner: constructs
src/agentic_test/cli/service_factory.py - Service Factory → Test Generation: constructs
src/agentic_test/cli/service_factory.py - Service Factory → Candidate Validation: constructs
src/agentic_test/cli/service_factory.py - Service Factory → Sandbox Execution: constructs
src/agentic_test/cli/service_factory.py - Service Factory → Failure Diagnosis: constructs
src/agentic_test/cli/service_factory.py - Service Factory → Workflow Engine: constructs
src/agentic_test/cli/service_factory.py - Workflow Engine → Workflow Nodes: dispatches
src/agentic_test/workflow/engine.py - Workflow Engine → Run Repository: persists checkpoints
src/agentic_test/workflow/engine.py - Workflow Engine → Event Store: emits events
src/agentic_test/workflow/engine.py - Workflow Engine → Workflow State: coordinates state
src/agentic_test/workflow/engine.py - Workflow Nodes → Repository Analysis: runs analysis
src/agentic_test/workflow/nodes.py - Workflow Nodes → Execution Planner: creates plan
src/agentic_test/workflow/nodes.py - Workflow Nodes → Test Generation: generates candidates
src/agentic_test/workflow/nodes.py - Workflow Nodes → Candidate Validation: validates candidates
src/agentic_test/workflow/nodes.py - Workflow Nodes → Sandbox Execution: runs sandbox
src/agentic_test/workflow/nodes.py - Workflow Nodes → Failure Diagnosis: diagnoses failures
src/agentic_test/workflow/nodes.py - Repository Analysis → Git Inspection: reads Git data
src/agentic_test/analysis/service.py - Repository Analysis → AST Analyzer: analyzes source
src/agentic_test/analysis/service.py - Repository Analysis → Test Discovery: discovers tests
src/agentic_test/analysis/service.py - Execution Planner → Test Discovery: uses test inventory
src/agentic_test/planning/planner.py - Execution Planner → Planning Rules: applies rules
src/agentic_test/planning/planner.py - Test Generation → Prompt Context: assembles prompts
src/agentic_test/generation/service.py - Prompt Context → LLM Adapter: requests completion
src/agentic_test/generation/context.py - Candidate Validation → Sandbox Execution: checks collection
src/agentic_test/validation/pipeline.py - Sandbox Execution → Coverage Metrics: collects metrics
src/agentic_test/execution/service.py - Failure Diagnosis → Triage Rules: classifies first
src/agentic_test/diagnosis/service.py - Failure Diagnosis → LLM Triage: falls back
src/agentic_test/diagnosis/service.py - LLM Triage → LLM Adapter: requests reasoning
src/agentic_test/diagnosis/llm_classifier.py - Run Repository → SQLite Database: reads and writes
src/agentic_test/storage/repository.py - Event Store → SQLite Database: stores events
src/agentic_test/storage/events.py - Workflow Recovery → Run Repository: loads checkpoints
src/agentic_test/workflow/recovery.py - Workflow Recovery → Event Store: reads run events
src/agentic_test/workflow/recovery.py