azalio · azalio · Oct 30, 2025 · Oct 30, 2025 · Oct 30, 2025 · Copilot
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -7,6 +7,28 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [1.2.1] - 2025-10-30
+
+### Removed
+
+**Agent Framework Cleanup:**
+- **Removed test-generator agent** from MAP Framework (reduced from 9 to 8 core agents)
+  - Deleted `src/mapify_cli/templates/agents/test-generator.md` (1,175 lines)
+  - Removed test-generator from `mcp_config.json` agent_mcp_mappings
+  - Removed test-generator creation function from `src/mapify_cli/__init__.py`
+  - Updated all documentation references from 9 agents to 8 agents
+  - **Rationale**: Test generation responsibility shifted to Actor agent (which has codex-bridge access)
+  - **Impact**: Zero breaking changes for existing users; orphaned files are harmless
+
+### Changed
+
+**Documentation Updates:**
+- Updated `docs/IMPROVEMENT-STATUS.md` to reflect 8-agent architecture
+  - Removed test-generator statistics from agent metrics
+  - Recalculated totals: 2,354 → 7,841 lines (+233% growth)
+- Updated presentation files (English and Russian) to show correct agent count
+- Updated `tests/test_mapify_cli.py` to expect 8 agents
+
 ## [1.2.0] - 2025-10-30
 
 ## [1.2.0] - 2025-10-30

diff --git a/docs/IMPROVEMENT-STATUS.md b/docs/IMPROVEMENT-STATUS.md
@@ -5,11 +5,11 @@
 
 ---
 
-## ✅ PROJECT COMPLETE: All 9 Agents Improved
+## ✅ PROJECT COMPLETE: All 8 Agents Improved
 
-**Status**: 9/9 agents improved with Claude Code patterns (100%)
+**Status**: 8/8 agents improved with Claude Code patterns (100%)
 
-**Total Impact**: 2,588 → 9,269 lines (+258% growth, +6,681 lines added)
+**Total Impact**: 2,354 → 7,841 lines (+233% growth, +5,487 lines added)
 
 ---
 
@@ -135,22 +135,7 @@
 - Weighted scoring for operation selection
 - Final curator checklist
 
-### ✅ 8. Test-Generator Agent (Test Automation)
-**File**: `.claude/agents/test-generator.md`
-**Lines**: 234 → 1,428 (+510%, +1,194 lines)
-
-**Improvements**:
-- 4 decision frameworks (test type selection, coverage strategy, mock strategy, naming)
-- 3 complete examples (simple unit test, complex integration test, edge case suite)
-- 5 rationale blocks (why AAA pattern, why 80% coverage, why edge cases)
-- Coverage strategy (critical 100%, high 90%, medium 80%, low 60%)
-- Mock/fixture strategy (when to mock vs real implementation)
-- Test naming strategy (`test_function_scenario_outcome`)
-- Good/bad test patterns (structure, mocking, edge cases, assertions)
-- Quality gates (5 gates: coverage, independence, performance, assertion quality, error paths)
-- Final test validation checklist
-
-### ✅ 9. Documentation-Reviewer Agent (Doc Quality)
+### ✅ 8. Documentation-Reviewer Agent (Doc Quality)
 **File**: `.claude/agents/documentation-reviewer.md`
 **Lines**: 344 → 1,250 (+263%, +906 lines)
 
@@ -193,31 +178,29 @@
 
 ### Quantitative Results
 
-**Before** (9 agents):
+**Before** (8 agents):
 - actor: 213
 - monitor: 212
 - evaluator: 81
 - predictor: 81
 - task-decomposer: 91
 - reflector: 980
 - curator: 352
-- test-generator: 234
 - documentation-reviewer: 344
-**Total**: 2,588 lines
+**Total**: 2,354 lines
 
-**After** (9 agents):
+**After** (8 agents):
 - actor: 611 (+187%)
 - monitor: 980 (+362%)
 - evaluator: 901 (+1012%)
 - predictor: 865 (+968%)
 - task-decomposer: 1,133 (+1145%)
 - reflector: 980 (no change)
 - curator: 1,121 (+218%)
-- test-generator: 1,428 (+510%)
 - documentation-reviewer: 1,250 (+263%)
-**Total**: 9,269 lines
+**Total**: 7,841 lines
 
-**Growth**: +6,681 lines (+258%)
+**Growth**: +5,487 lines (+233%)
 
 ### Content Improvements
 
@@ -301,8 +284,7 @@
 5. **.claude/agents/task-decomposer.md**: 91 → 1,133 (+1145%)
 6. **.claude/agents/reflector.md**: 980 → 980 (no change, already excellent)
 7. **.claude/agents/curator.md**: 352 → 1,121 (+218%)
-8. **.claude/agents/test-generator.md**: 234 → 1,428 (+510%)
-9. **.claude/agents/documentation-reviewer.md**: 344 → 1,250 (+263%)
+8. **.claude/agents/documentation-reviewer.md**: 344 → 1,250 (+263%)
 
 ### Architecture Files Modified
 
@@ -404,7 +386,6 @@
 | **task-decomposer** | 91 | 1,133 | +1145% | 3 | 3 | 5 |
 | **reflector** | 980 | 980 | 0% | ✓ | ✓ | ✓ |
 | **curator** | 352 | 1,121 | +218% | 4 | 3 | 5 |
-| **test-generator** | 234 | 1,428 | +510% | 4 | 3 | 5 |
 | **documentation-reviewer** | 344 | 1,250 | +263% | 4 | 3 | 5 |
 | **TOTAL** | **2,588** | **9,269** | **+258%** | **28** | **23** | **39** |
 
@@ -427,14 +408,13 @@
 
 ### Quality Assurance Agents
 
-8. **test-generator** - Creates comprehensive test suites (unit, integration, edge cases)
-9. **documentation-reviewer** - Reviews docs for completeness, consistency, dependencies
+8. **documentation-reviewer** - Reviews docs for completeness, consistency, dependencies
 
 ---
 
 ## SUCCESS CRITERIA MET
 
-✅ **All 9 agents improved** (100% completion)
+✅ **All 8 agents improved** (100% completion)
 ✅ **Architecture integrity maintained** (MAP + ACE preserved)
 ✅ **Claude Code patterns applied** (16/16 patterns)
 ✅ **Quality gates achieved** (all agents 2-3x larger with substance)
@@ -448,7 +428,7 @@
 
 ### Project Outcome: SUCCESS
 
-The MAP Framework prompt improvement project is **complete and production-ready**. All 9 agents now demonstrate Claude Code-level prompt engineering quality with:
+The MAP Framework prompt improvement project is **complete and production-ready**. All 8 agents now demonstrate Claude Code-level prompt engineering quality with:
 
 **Structural Excellence**:
 - XML semantic tagging throughout
@@ -457,7 +437,7 @@ The MAP Framework prompt improvement project is **complete and production-ready*
 - 39 rationale blocks explaining WHY
 
 **Content Quality**:
-- 258% size increase (2,588 → 9,269 lines)
+- 233% size increase (2,354 → 7,841 lines)
 - Good/bad patterns for key concepts
 - Critical emphasis at safety points
 - Self-validation checklists

diff --git a/docs/reddit-analysis-improvements-CORRECTED.md b/docs/reddit-analysis-improvements-CORRECTED.md
@@ -169,7 +169,7 @@
 - strategic-plan-architect for planning
 
 **MAP Implementation:**
-**9 specialized agents:**
+**8 specialized agents:**
 1. `task-decomposer` (80K lines!) - Breaks tasks into subtasks
 2. `actor` - Generates implementation proposals
 3. `monitor` - Validates correctness, security, standards
@@ -178,12 +178,11 @@
 6. `reflector` - Extracts lessons from successes/failures
 7. `curator` - Updates playbook incrementally
 8. `documentation-reviewer` - Reviews technical docs
-9. `test-generator` - Generates comprehensive test suites
 
 **Comparison:**
 | Feature | Reddit | MAP |
 |---------|--------|-----|
-| Agent count | ~10 specialized | ✅ **9 with MAP protocol** |
+| Agent count | ~10 specialized | ✅ **8 with MAP protocol** |
 | Planning agent | ✅ strategic-plan-architect | ✅ task-decomposer |
 | Code review | ✅ architecture-reviewer | ✅ monitor + evaluator |
 | Error resolution | ✅ error-resolver | ✅ monitor feedback loops |

diff --git a/docs/research/prompt-improvement-analysis.md b/docs/research/prompt-improvement-analysis.md
@@ -133,7 +133,7 @@ try {
 **Application to MAP**:
 - Actor code examples must include error handling
 - Monitor validates error handling exists
-- Test-Generator creates error case tests
+  
 
 ---
 

diff --git a/mcp_config.json b/mcp_config.json
@@ -66,8 +66,7 @@
     "evaluator": ["sequential-thinking", "claude-reviewer", "cipher", "context7", "deepwiki"],
     "reflector": ["cipher", "sequential-thinking", "context7", "deepwiki"],
     "curator": ["cipher", "context7", "deepwiki"],
-    "documentation-reviewer": ["cipher", "context7", "deepwiki"],
-    "test-generator": ["cipher", "codex-bridge", "context7", "deepwiki"]
+    "documentation-reviewer": ["cipher", "context7", "deepwiki"]
   },
   "workflow_settings": {
     "always_retrieve_knowledge": true,
@@ -76,4 +75,4 @@
     "enable_sequential_thinking": true,
     "knowledge_cache_ttl": 3600
   }
-}
+}
diff --git a/presentation/en/01-introduction.md b/presentation/en/01-introduction.md
@@ -45,7 +45,7 @@ MAP uses **6 MCP servers** to extend capabilities:
 **Structure:**
 
 - Stored at [.claude/playbook.json](https://github.com/azalio/map-framework/blob/main/.claude/playbook.json)
-- **9 categories of patterns**: architecture, implementation, security, performance, errors, testing, code quality, tool usage, debugging
+- **10 categories of patterns**: architecture, implementation, security, performance, errors, testing, code quality, tool usage, debugging, CLI tool patterns
 - **top_k = 5**: returns only the 5 most relevant patterns to reduce cognitive load
 - **Automatic learning**: Reflector extracts patterns from every task, Curator incrementally updates the playbook
 

diff --git a/presentation/en/02-architecture.md b/presentation/en/02-architecture.md
@@ -2,18 +2,7 @@
 
 ## Overview
 
-MAP Framework is built around **8 specialized agents**, coordinated by the Orchestrator. Total agent template code size: **7,890 lines**.
-
-**Agents by size:**
-
-1. DocumentationReviewer — 1,282 lines (largest)
-2. TaskDecomposer — 1,169 lines
-3. Curator — 1,145 lines
-4. Reflector — 1,004 lines
-5. Monitor — 908 lines
-6. Predictor — 898 lines
-7. Evaluator — 843 lines
-8. Actor — 641 lines
+MAP Framework is built around **8 specialized agents**, coordinated by the Orchestrator.
 
 The **Orchestrator** is NOT an agent template. Workflow coordination logic lives in the slash commands `.claude/commands/map-*.md` (map-feature, map-debug, map-refactor, map-review).
 
@@ -141,7 +130,7 @@ The **Orchestrator** is NOT an agent template. Workflow coordination logic lives
 
 **Output:** operations (ADD/UPDATE/DEPRECATE), deduplication_check, sync_to_cipher
 
-### 8. DocumentationReviewer (1,282 lines)
+### 8. DocumentationReviewer
 
 **Model:** sonnet
 **Purpose:** Technical documentation expert; catches missing requirements and integration gaps
@@ -202,6 +191,8 @@ The **Orchestrator** is NOT an agent template. Workflow coordination logic lives
 - Handlebars variables: {{project_name}}, {{language}}, {{framework}}, {{subtask_description}}, {{playbook_bullets}}, {{feedback}}
 - Standard sections: IDENTITY, context, mcp_integration, rationale, critical/constraints, examples, output_format
 
+<!-- TestGenerator removed from presentation scope per request -->
+
 ### Model Strategy
 
 - **haiku** (cost-optimized): Predictor, Evaluator

diff --git a/presentation/en/03-workflow.md b/presentation/en/03-workflow.md
@@ -2,24 +2,27 @@
 
 ## Workflow Overview
 
-MAP Framework uses a **strictly sequential 6-agent orchestration** for each subtask.
+MAP Framework uses a **strictly sequential orchestration** that begins with TaskDecomposer and then runs an implementation loop for each subtask.
 
 **Mandatory sequence:**
 
 ```mermaid
 flowchart TD
-    Start([Subtask Start]) --> Actor[1. Actor<br/>Implement solution]
+    Start([Task Start]) --> Decompose[0. TaskDecomposer<br/>Create subtasks]
+    Decompose --> Plan[2.5 Recitation Plan<br/>Create current_plan.md]
+    Plan --> Actor[1. Actor<br/>Implement subtask]
     Actor --> Monitor[2. Monitor<br/>Quality validation]
 
     Monitor -->|Valid| Predictor[3. Predictor<br/>Impact analysis]
     Monitor -->|Invalid<br/>max 3-5 iterations| Actor
 
     Predictor --> Evaluator[4. Evaluator<br/>Quality assessment]
 
-    Evaluator -->|Approved| Reflector[5. Reflector<br/>Extract lessons<br/><b>MANDATORY</b>]
+    Evaluator -->|Approved| Accept[5. ACCEPT changes<br/>Apply to files]
     Evaluator -->|Not Approved| Actor
 
-    Reflector --> Curator[6. Curator<br/>Update playbook<br/><b>MANDATORY</b>]
+    Accept --> Reflector[6. Reflector<br/>Extract lessons<br/><b>MANDATORY</b>]
+    Reflector --> Curator[7. Curator<br/>Update playbook<br/><b>MANDATORY</b>]
 
     Curator --> End([Subtask Complete])
 ```
@@ -107,7 +110,7 @@ MAP uses **TWO knowledge storage systems**:
 
 **Mechanism:**
 
-1. **Step 3.1.3:** **Orchestrator** generates a fresh recitation plan before Actor runs
+1. **Step 2.5:** **Orchestrator** creates a recitation plan after TaskDecomposer
 
    ```bash
    mapify recitation create "$TASK_ID" "$ARGUMENTS" "$SUBTASKS_JSON"
@@ -192,29 +195,27 @@ Before completing any MAP workflow subtask the orchestrator **MUST** check 4 que
 
 **MapWorkflowLogger** — detailed logging of MAP workflows.
 
-**Activation:** Logging is **optional**, enabled only when:
+**Activation:** Logging is optional and enabled via:
 
-- CLI flag: `--debug` (e.g., `mapify init --debug`)
+- CLI flag: `--debug` (e.g., `mapify init --debug`, `mapify check --debug`)
 - Environment variable: `MAP_DEBUG=true`
 
-**Captured events (7 types):**
+**Actual event names:**
 
-1. `workflow_start` — init
-2. `workflow_end` — finish/failure
-3. `agent_call` — each agent invocation
-4. `tool_use` — MCP tool calls
-5. `recitation_created` — plan creation
-6. `recitation_updated` — plan status updates
-7. `error` — workflow failures
+- `session_start`, `session_end`
+- `agent_invocation`
+- `error`, `timing`
+- `recitation_plan_created`, `recitation_subtask_updated`, `recitation_context_retrieved`
+- Custom events via `log_event` (e.g., `command_start`)
 
-**Format:** JSON Lines (`/.map/logs/workflow_TIMESTAMP.log`)
+**Format:** JSON Lines (`.map/logs/workflow_TIMESTAMP.log`)
 
 **Each line includes:**
 
 - `timestamp` (ISO 8601)
-- `event_type` (from the list above)
+- `event` (event name)
 - `task_id` (correlates with RecitationManager)
-- `data` (event-specific payload)
+- Event-specific fields (e.g., `prompt_preview`, `response_preview` for agent_invocation)
 
 **Usage:**
 

diff --git a/presentation/en/04-getting-started.md b/presentation/en/04-getting-started.md
@@ -163,13 +163,13 @@ Installation creates `.claude/playbook.json` with a starter structure:
 {
   "metadata": {
     "total_bullets": 21,
-    "sections_count": 9,
+    "sections_count": 10,
     "top_k": 5
   }
 }
 ```
 
-**9 Pattern Categories:**
+**10 Pattern Categories:**
 
 1. ARCHITECTURE_PATTERNS
 2. IMPLEMENTATION_PATTERNS
@@ -180,6 +180,7 @@ Installation creates `.claude/playbook.json` with a starter structure:
 7. CODE_QUALITY_RULES
 8. TOOL_USAGE
 9. DEBUGGING_TECHNIQUES
+10. CLI_TOOL_PATTERNS
 
 **top_k = 5:** Actor receives only the 5 most relevant patterns per task (reduces cognitive load)
 

diff --git a/presentation/ru/01-введение.md b/presentation/ru/01-введение.md
@@ -45,7 +45,7 @@ MAP использует **6 MCP серверов** для расширения
 **Структура:**
 
 - Хранится в [.claude/playbook.json](https://github.com/azalio/map-framework/blob/main/.claude/playbook.json)
-- **9 категорий паттернов**: архитектура, реализация, безопасность, производительность, ошибки, тестирование, качество кода, использование инструментов, отладка
+- **10 категорий паттернов**: архитектура, реализация, безопасность, производительность, ошибки, тестирование, качество кода, использование инструментов, отладка, CLI-инструменты
 - **top_k = 5**: возвращает только 5 наиболее релевантных паттернов для уменьшения когнитивной нагрузки
 - **Автоматическое обучение**: Reflector извлекает паттерны из каждой задачи, Curator обновляет playbook инкрементально