AI Agent MCP Server

# Progress Steps ## CLAUDE.md - [X] Move researches to artifacts/researches and update references - [X] Rename product_vision_v1.md to proper naming format, move it to product_visions, and update references - [X] File/folder structures for artifacts - [X] Revise sub TODO.md (tracking) - [X] Verify the references of those TODO.md files - [X] Drop sub TODO.md files ## Validation - [X] Move tracebility generator validation checks to a proper section - [X] Review generators for paths in tracebility sections - [X] Generator output file paths - [X] Read all generated epics - [X] Read all restructured research docs (business and implementation) - [X] Read all generators and templates (file paths removed, no validation duplications, defragmented validation instructions) - [X] Check output file name for generators - [X] Remove all paths from generators and templates. Keep it only in CLAUDE.md (input, output) - [X] Evaluation related to required/optional input data for generators - [X] Remove Tracibility Checklists from templates (validate existance in generators) - [X] verify generate and refine prompts - [X] examples (Step 5, Step 6 -> Updated Sections) - [X] create refinemenets plan for generated artifacts and apply - [X] apply Issue 1, PRD gen only - [X] apply Issue 2, PRD only - [X] apply Issue 3, all - [X] Validate that everything was applied - [X] CLAUDE.md consolidation after generate and refine prompts refinement - [X] Phase 1 (Revised): Removed entire "Artifact Path Patterns" section (100% redundancy elimination) - [X] Phase 2: Simplified SDLC Artifact Dependency Flow (removed path redundancy) - [X] Phase 3: Merged File Naming Conventions into Folder Structure (single source of truth) - [X] Manual: Update 12 generator prompts to reference "Folder Structure" patterns - [X] Open Questions in EPIC-000 too techy ## FOR EVERY NEW PRD: Validation - [ ] Technical Consideration sections were removed from Epics and saved into `artifacts/technical_considerations_from_epic.md` - [ ] Validate that these sections are generated in PRDs and compare the content - [ ] Open Questions (technical) from all epics -> PRDs ## Semi-automatic testing - [ ] feedback/context_engineering_strategy_v1_critique_v3.md - [ ] rewrite TASK_003 based on critique_v3 file before execution - [ ] Review execute-generator - [ ] verify Claude context is in the task report (execute-generator) - [ ] Review refine ## Pre-finalization - [ ] `docs/research/advanced_prompt_engineering/research_coverage_evaluation.md` - [X] verify that all requests from CHAT.md are applied (or create AI report) - [X] review all documents from the beginning - [ ] analyze all `claude-md` documents - [ ] review strategy document, if is still needed ## Finalization - [X] remove CLAUDE-*.md lazy generation instructions, keep "if exists" verification - [X] remove next-generator instructions, keep only reference to next-generator to be invoked in the report - [ ] Validation checks migration to a validation tool - [ ] strategy doc update - [ ] GRADUATION.md review - [ ] IMPORTANT: compare with dot-ai, prp, portal projects from sandbox ## Implementation ### Where is the specification/guidelines on: - [ ] entities - [ ] value objects - [ ] data models - [ ] API contracts to be created - [ ] DTO - [ ] interfaces - [ ] abstract classes - [ ] patterns (observability, security, encryption, logging ...) - [ ] libraries (utility/common, external integration) - [ ] frameworks / SDKs ### QA Test Plan test plan creation ## Next Phases ### Phase 2 * software implementation generator prompts (source code, unit tests, deployments) * specialed CLAUDE.md files in `claude.md` project * TODO.md tasks * strategy doc update * GRADUATION.md ### Phase 3 * generate MCP Server implementation * TODO.md tasks * strategy doc update ### Phase 4 * move AI artifacts to MCP server * tracking strategy (tracking TODO.md files, PRDs, USs ...) * strategy doc update * add new product repo initialization tool to MCP Server ### Phase 5 * Backlog product * make all product management artifact related and trackable * PRD has multiple sections, each sections should be linked to something within sub-artifact (Functional req from PRD is linked to functional req of user story). This gives a clear pictures of general requirement coverage (analogus to Code coverage with tests) * Backlog event should trigger re-indexing of document repository items within RAG system * adding of a new User Story, can add new feature in PRD, which shell trigger PRD re-index ### Phase 6 * SHH tool using MCP Server (final MCP server end-to-end test) * add full git support (PR merge requests) * add CI/CD ### Phase 7 * add RAG tool to MCP server * every RAG indexed item should have re-index trigger defined to prohibit out-of-sync RAG results * index all AI artifacts * index source code ### Phase 8 * MCP server graduation to semi-automation/full automation software implementation * Task orchestration automatization with Backlog tasks * Sub-agents