CHERENKOV Development Plan¶
This document breaks down the development phases into actionable sprints for AI agents and human developers.
Guiding Principle¶
Test-Driven Development (TDD) is mandatory. Write the test, run the test (it should fail), write the code, run the test (it should pass), then refactor.
Sprint 1: Formalize Scanners (Current)¶
Goal: Migrate candidate scanners to a uniform interface.
1. Task: Define BaseScanner abstract class in src/cherenkov/scanners/base.py.
2. Task: Refactor cherenkov_simple_scanner.py to inherit from BaseScanner.
3. Task: Implement unit tests for the base interface ensuring strict input/output validation.
4. Autonomy Level: Fully autonomous.
Sprint 2: Parallel Orchestration Core¶
Goal: Refactor the orchestrator to support concurrent execution.
1. Task: Update ai_workflows_orchestrator.py to use asyncio for executing agent tasks concurrently.
2. Task: Implement AIMD circuit breaker logic for agent calls.
3. Task: Write integration tests proving parallel execution finishes faster than sequential execution without errors.
4. Autonomy Level: Fully autonomous.
Sprint 3: The Validation Sandbox (TOKAMAK)¶
Goal: Implement the isolated execution environment for PoCs.
1. Task: Create Docker definitions for the TOKAMAK sandbox (minimal, no network).
2. Task: Implement the Command pattern for sending payloads to the sandbox.
3. Task: Implement cryptographic signing of the execution results (BurhanTrace).
4. Autonomy Level: Human Review Required (Security Architecture).
Sprint 4: Human-in-the-Loop (HITL) Workflows¶
Goal: Ensure critical actions cannot be performed without human consent. 1. Task: Implement an API endpoint to pause workflow execution pending approval. 2. Task: Update UI to present pending approvals. 3. Task: Implement RBAC to verify the approving user has sufficient clearance. 4. Autonomy Level: Human Review Required (Core Security Feature).
Sprint 5: Compliance and Reporting¶
Goal: Map findings to regulatory frameworks. 1. Task: Build the compliance mapping engine (CVE/CWE -> SAMA/EGY-FIN). 2. Task: Integrate local PDF generation for audit reports. 3. Autonomy Level: Fully autonomous.