tachibot-mcp 2.0.6 → 2.1.0

package/docs/API_KEYS.md

@@ -48,7 +48,7 @@ TachiBot MCP works with multiple AI providers to offer diverse capabilities. You
 |----------|--------------|----------------|
 | **Perplexity** | Research, web search | `perplexity_ask`, `perplexity_research`, `perplexity_reason`, `scout` (default) |
 | **Grok/xAI** | Live search, reasoning | `grok_search`, `grok_reason`, `grok_code`, `grok_debug`, `grok_architect`, `grok_brainstorm`, `scout` (with grok) |
-| **OpenAI** | GPT-5 models | `openai_brainstorm`, `openai_compare`, `focus` (some modes), `verifier`, `challenger` |
+| **OpenAI** | GPT-5 models | `openai_brainstorm`, `openai_reason`, `openai_code_review`, `openai_explain`, `focus` (some modes), `verifier`, `challenger` |
 | **Google** | Gemini models | `gemini_brainstorm`, `gemini_analyze_code`, `gemini_analyze_text`, `verifier`, `scout` |
 | **OpenRouter** | Qwen models | `qwen_coder`, `qwen_competitive` |
 
@@ -116,11 +116,11 @@ Grok (by xAI) provides live web search, reasoning, and code analysis.
 
 #### Models Available
 
-- **grok-4.1** - Latest (Nov 2025): Enhanced reasoning, creativity & emotional intelligence
-- **grok-4.1-fast** - Tool-calling optimized: Fast inference, agentic workflows
-- **grok-4** - Previous reasoning model
-- **grok-4-0709** - Specific version
-- **grok-4-heavy** - Extended context (256k tokens)
+- **grok-4-1-fast-reasoning** - Latest (Nov 2025): Enhanced reasoning, creativity & emotional intelligence (2M context)
+- **grok-4-1-fast-non-reasoning** - Tool-calling optimized: Fast inference, agentic workflows (2M context)
+- **grok-4-fast-reasoning** - Previous reasoning model
+- **grok-4-0709** - Heavy model (expensive, use sparingly)
+- **grok-code-fast-1** - Coding specialist
 
 #### Pricing
 
@@ -175,18 +175,21 @@ OpenAI provides GPT-5 models for brainstorming, comparison, and reasoning.
 
 #### Models Available
 
-- **gpt-5** - Full GPT-5 with deep reasoning
-- **gpt-5-mini** - Faster, cheaper GPT-5
-- **gpt-5-nano** - Quickest, cheapest
-- **qwq-32b**, **qwen3-30b**, **qwen3-coder-480b** - Via OpenAI (if configured)
+- **gpt-5.1** - Flagship model with deep reasoning (2M context)
+- **gpt-5.1-codex-mini** - Fast, cheap workhorse for code tasks (256K context)
+- **gpt-5.1-codex** - Power model for complex code (1M context)
+- **gpt-5-pro** - Premium for complex orchestration (4M context)
 
 #### Pricing
 
+> **Note:** Prices are approximate and may be outdated. Check [OpenAI Pricing](https://openai.com/pricing) for current rates.
+
 | Model | Input | Output | Notes |
 |-------|-------|--------|-------|
-| GPT-5 | $15.00 / 1M tokens | $60.00 / 1M tokens | Full reasoning |
-| GPT-5-mini | $2.00 / 1M tokens | $8.00 / 1M tokens | Most balanced |
-| GPT-5-nano | $0.40 / 1M tokens | $1.60 / 1M tokens | Budget option |
+| gpt-5.1 | ~$10 / 1M tokens | ~$30 / 1M tokens | Flagship reasoning |
+| gpt-5.1-codex-mini | ~$2 / 1M tokens | ~$6 / 1M tokens | Best value for code |
+| gpt-5.1-codex | ~$15 / 1M tokens | ~$45 / 1M tokens | Complex code tasks |
+| gpt-5-pro | ~$20 / 1M tokens | ~$60 / 1M tokens | Premium orchestration |
 
 **Warning:** GPT-5 models may generate invisible reasoning tokens that increase costs. Monitor usage carefully.
 
@@ -205,11 +208,11 @@ OpenAI provides GPT-5 models for brainstorming, comparison, and reasoning.
 
 #### Cost Estimation
 
-- Single `openai_brainstorm` (gpt-5-mini): ~$0.01 - $0.03
+- Single `openai_brainstorm` (gpt-5.1-codex-mini): ~$0.01 - $0.03
 - Single `openai_brainstorm` (gpt-5): ~$0.15 - $0.40
-- Single `openai_compare`: ~$0.02 - $0.05
+- Single `openai_code_review`: ~$0.02 - $0.05
 
-**Tip:** Use `model: "gpt-5-mini"` by default, only use `gpt-5` for complex tasks.
+**Tip:** Use `model: "gpt-5.1-codex-mini"` by default, only use `gpt-5` for complex tasks.
 
 #### Add to .env
 
@@ -313,6 +316,37 @@ Varies by model, generally:
 OPENROUTER_API_KEY=sk-or-v1-abc123...
 ```
 
+#### OpenRouter Gateway Mode (Optional)
+
+OpenRouter can act as a **unified gateway** for all providers (OpenAI, Gemini, Grok) with a single API key:
+
+```bash
+# Enable gateway mode - routes all providers through OpenRouter
+USE_OPENROUTER_GATEWAY=true
+OPENROUTER_API_KEY=sk-or-v1-abc123...
+```
+
+**How it works:**
+| Provider | Default Mode | Gateway Mode |
+|----------|--------------|--------------|
+| Kimi/Qwen | OpenRouter | OpenRouter (no change) |
+| OpenAI | Direct API | → OpenRouter |
+| Gemini | Direct API | → OpenRouter |
+| Grok | Direct API | → OpenRouter |
+| Perplexity | Direct API | Direct API (always) |
+
+**Benefits:**
+- ✅ Single API key for most providers
+- ✅ Unified billing dashboard
+- ✅ Automatic fallback/load balancing
+
+**Limitations:**
+- ⚠️ Perplexity still requires direct API (not on OpenRouter)
+- ⚠️ Some provider-specific features may not work (e.g., `reasoning_effort`)
+- ⚠️ Slight latency overhead (proxy)
+
+**Note:** Gateway mode is validated by Andrej Karpathy's [llm-council](https://github.com/karpathy/llm-council) project.
+
 ---
 
 ## Cost Comparison
@@ -434,7 +468,7 @@ See [TOOL_PROFILES.md](TOOL_PROFILES.md) for details.
 - **Deep research:** `perplexity_research` (expensive, use sparingly)
 - **Live data:** `grok_search` with low `maxSearchSources` (10-20)
 - **Code tasks:** `gemini_analyze_code` or `qwen_coder` (cost-effective)
-- **Brainstorming:** `gemini_brainstorm` or `openai_brainstorm` with `model: "gpt-5-mini"`
+- **Brainstorming:** `gemini_brainstorm` or `openai_brainstorm` with `model: "gpt-5.1-codex-mini"`
 
 ### 4. Monitor Regularly
 
@@ -509,7 +543,7 @@ TACHI_CACHE_TTL=3600  # 1 hour
 2. Review which tools are being used
 3. Switch to `minimal` or `balanced` profile
 4. Avoid `grok_search` with high `maxSearchSources`
-5. Use `gpt-5-mini` instead of `gpt-5`
+5. Use `gpt-5.1-codex-mini` instead of `gpt-5`
 
 ### API Key Not Working After Setup
 

package/docs/CONFIGURATION.md

@@ -167,6 +167,23 @@ ANTHROPIC_API_KEY=sk-ant-...
 QWEN_API_KEY=...
 ```
 
+### OpenRouter Gateway Mode
+
+Route all providers (OpenAI, Gemini, Grok) through OpenRouter with a single API key:
+
+```bash
+# Enable gateway mode
+USE_OPENROUTER_GATEWAY=true
+OPENROUTER_API_KEY=sk-or-...
+```
+
+**Routing behavior:**
+- **Kimi/Qwen** → Always OpenRouter (native)
+- **OpenAI/Gemini/Grok** → Direct API (default) or OpenRouter (when gateway enabled)
+- **Perplexity** → Always direct API (not on OpenRouter)
+
+See [API_KEYS.md](API_KEYS.md#openrouter-gateway-mode-optional) for details.
+
 ### Search Configuration
 
 ```bash
@@ -249,28 +266,28 @@ Configure which models are used for Scout, Challenger, and Verifier tools. These
 
 ```bash
 # Scout model configuration
-SCOUT_QUICK_MODELS=qwen/qwen3-coder-plus,gemini-2.5-flash,gpt-5-mini
-SCOUT_RESEARCH_MODELS=qwen/qwen3-coder-plus,gemini-2.5-pro,gpt-5-mini
+SCOUT_QUICK_MODELS=qwen/qwen3-coder-plus,gemini-2.5-flash,gpt-5.1-codex-mini
+SCOUT_RESEARCH_MODELS=qwen/qwen3-coder-plus,gemini-2.5-pro,gpt-5.1-codex-mini
 
 # Challenger model configuration
-CHALLENGER_MODELS=qwen/qwen3-coder-plus,gemini-2.5-pro,gpt-5-mini
+CHALLENGER_MODELS=qwen/qwen3-coder-plus,gemini-2.5-pro,gpt-5.1-codex-mini
 
 # Verifier model configuration
-VERIFIER_QUICK_MODELS=qwen/qwen3-coder-plus,gemini-2.5-flash,gpt-5-mini
-VERIFIER_STANDARD_MODELS=qwen/qwen3-coder-plus,gemini-2.5-pro,gpt-5-mini
+VERIFIER_QUICK_MODELS=qwen/qwen3-coder-plus,gemini-2.5-flash,gpt-5.1-codex-mini
+VERIFIER_STANDARD_MODELS=qwen/qwen3-coder-plus,gemini-2.5-pro,gpt-5.1-codex-mini
 VERIFIER_DEEP_MODELS=qwen/qwen3-coder-plus,gemini-2.5-pro,gpt-5
 
 # Default models for fallback
-DEFAULT_MODELS=qwen/qwen3-coder-plus,gemini-2.5-pro,gpt-5-mini
+DEFAULT_MODELS=qwen/qwen3-coder-plus,gemini-2.5-pro,gpt-5.1-codex-mini
 ```
 
 **Cost Optimization:**
-- **gpt-5-mini**: 60% cheaper (~$0.50/$1.00 per 1M tokens), faster, good for most tasks
+- **gpt-5.1-codex-mini**: 60% cheaper (~$0.50/$1.00 per 1M tokens), faster, good for most tasks
 - **gpt-5**: Full quality (~$1.25/$2.50 per 1M tokens), best for critical decisions
 - **gemini-2.5-flash**: Faster, cheaper, good for quick checks
 - **gemini-2.5-pro**: Better reasoning/accuracy, recommended for verification
 
-**Recommendation:** Use defaults (gpt-5-mini) for 60% cost savings. Upgrade to gpt-5 only for `deep_verify` or critical workflows.
+**Recommendation:** Use defaults (gpt-5.1-codex-mini) for 60% cost savings. Upgrade to gpt-5 only for `deep_verify` or critical workflows.
 
 ### Model-Specific Settings
 

package/docs/TOOLS_REFERENCE.md

@@ -22,9 +22,8 @@
   - [grok_architect](#grok_architect)
   - [grok_brainstorm](#grok_brainstorm)
 - [OpenAI Suite](#openai-suite)
-  - [openai_gpt5_reason](#openai_gpt5_reason)
+  - [openai_reason](#openai_reason)
   - [openai_brainstorm](#openai_brainstorm)
-  - [openai_compare](#openai_compare)
   - [openai_code_review](#openai_code_review)
   - [openai_explain](#openai_explain)
 - [Gemini Suite](#gemini-suite)
@@ -837,7 +836,7 @@ Creative brainstorming using GPT-5 suite with advanced controls.
 ```typescript
 {
   problem: string;                  // REQUIRED
-  model?: "gpt-5" | "gpt-5-mini" | "gpt-5-nano";  // Default: "gpt-5-mini"
+  model?: "gpt-5.1" | "gpt-5.1-codex-mini" | "gpt-5.1-codex";  // Default: "gpt-5.1-codex-mini"
   quantity?: number;                // Default: 5
   style?: "innovative" | "practical" | "wild" | "systematic";
   constraints?: string;
@@ -852,7 +851,7 @@ Creative brainstorming using GPT-5 suite with advanced controls.
 | Parameter | Type | Required | Default | Description |
 |-----------|------|----------|---------|-------------|
 | `problem` | `string` | ✅ Yes | - | Problem to brainstorm |
-| `model` | `string` | No | `"gpt-5-mini"` | GPT-5 model variant |
+| `model` | `string` | No | `"gpt-5.1-codex-mini"` | GPT-5 model variant |
 | `quantity` | `number` | No | `5` | Number of ideas to generate |
 | `style` | `string` | No | `"innovative"` | Brainstorming style |
 | `constraints` | `string` | No | - | Additional constraints |
@@ -864,9 +863,9 @@ Creative brainstorming using GPT-5 suite with advanced controls.
 
 | Model | Speed | Cost | Best For |
 |-------|-------|------|----------|
-| `gpt-5-nano` | Fastest | $ | Quick ideation |
-| `gpt-5-mini` | Fast | $$ | Most tasks (default) |
-| `gpt-5` | Slow | $$$$ | Complex problems requiring deep reasoning |
+| `gpt-5.1-codex-mini` | Fast | $$ | Most tasks (default) |
+| `gpt-5.1-codex` | Medium | $$$ | Complex code tasks |
+| `gpt-5.1` | Slow | $$$$ | Deep reasoning problems |
 
 #### Reasoning Effort (GPT-5 only)
 
@@ -909,7 +908,7 @@ openai_brainstorm({
 ```typescript
 openai_brainstorm({
   problem: "Reduce app cold start time",
-  model: "gpt-5-nano",
+  model: "gpt-5.1-codex-mini",
   quantity: 5,
   style: "practical",
   constraints: "Must work on mobile devices"
@@ -928,41 +927,6 @@ openai_brainstorm({
 
 ---
 
-### openai_compare
-
-Multi-option consensus analysis with GPT-5.
-
-#### Schema
-
-```typescript
-{
-  topic: string;                    // REQUIRED
-  options: string[];                // REQUIRED - Options to compare
-  criteria?: string[];              // Evaluation criteria
-  includeRecommendation?: boolean;  // Default: true
-}
-```
-
-#### Example Calls
-
-**Compare frameworks:**
-```typescript
-openai_compare({
-  topic: "JavaScript framework selection",
-  options: ["React", "Vue", "Svelte", "Angular"],
-  criteria: [
-    "Learning curve",
-    "Performance",
-    "Community support",
-    "Ecosystem maturity",
-    "Job market demand"
-  ],
-  includeRecommendation: true
-})
-```
-
----
-
 ## Gemini Suite
 
 ### gemini_brainstorm
@@ -1155,7 +1119,7 @@ Multi-model parallel verification with consensus analysis.
 #### Variants
 
 **quick_verify** (Default)
-- Models: `gpt-5-mini`, `gemini-2.5-flash`, `gpt-5`
+- Models: `gpt-5.1-codex-mini`, `gemini-2.5-flash`, `gpt-5`
 - Tokens: 2000
 - Timeout: 10s
 - Use: Fast verification
@@ -1167,7 +1131,7 @@ Multi-model parallel verification with consensus analysis.
 - Use: Complex reasoning
 
 **fact_check**
-- Models: `gpt-5`, `gemini-2.5-pro`, `gpt-5-mini`
+- Models: `gpt-5`, `gemini-2.5-pro`, `gpt-5.1-codex-mini`
 - Tokens: 3000
 - Timeout: 15s
 - Sources: Enabled by default
@@ -1339,7 +1303,7 @@ Critical thinking and echo chamber prevention by generating counter-arguments.
 ```typescript
 {
   context: string | object | array;  // REQUIRED
-  model?: string;                    // Default: "gpt-5-mini"
+  model?: string;                    // Default: "gpt-5.1-codex-mini"
   maxTokens?: number;                // Default: 2000
   temperature?: number;              // 0-1, Default: 0.9
 }
@@ -1350,13 +1314,13 @@ Critical thinking and echo chamber prevention by generating counter-arguments.
 | Parameter | Type | Required | Default | Description |
 |-----------|------|----------|---------|-------------|
 | `context` | `string \| object \| array` | ✅ Yes | - | Claims to challenge |
-| `model` | `string` | No | `"gpt-5-mini"` | AI model to use |
+| `model` | `string` | No | `"gpt-5.1-codex-mini"` | AI model to use |
 | `maxTokens` | `number` | No | `2000` | Max tokens per call |
 | `temperature` | `number` | No | `0.9` | Creativity (0-1) |
 
 #### Supported Models
 
-- `gpt-5-mini`, `gpt-5`, `qwq-32b`, `qwen3-30b`, `qwen3-coder-480b`
+- `gpt-5.1-codex-mini`, `gpt-5`, `qwq-32b`, `qwen3-30b`, `qwen3-coder-480b`
 - `gemini-2.5-flash`, `gemini-2.5-pro`
 - `grok-4`, `grok-4-0709`
 - `sonar-pro`, `perplexity-sonar-pro`

package/docs/TOOL_PARAMETERS.md

@@ -20,7 +20,7 @@ The Challenger tool provides critical thinking and echo chamber prevention by ge
 | Parameter | Type | Required | Default | Description |
 |-----------|------|----------|---------|-------------|
 | `context` | `string \| object \| array` | ✅ Yes | - | The claims, statements, or context to challenge. Can be a string, object with `query`/`text`/`content`, or array of contexts |
-| `model` | `string` | No | `'gpt-5-mini'` | AI model to use for generating challenges. Supports: `gpt-5-mini`, `gpt-5`, `gemini-2.5-flash`, `gemini-2.5-pro`, `grok-4`, `sonar-pro` |
+| `model` | `string` | No | `'gpt-5.1-codex-mini'` | AI model to use for generating challenges. See [Supported Models](#supported-models) section |
 | `maxTokens` | `number` | No | `2000` | Maximum tokens per API call |
 | `temperature` | `number` | No | `0.9` | Temperature for response generation (0-1). Higher = more creative challenges |
 
@@ -89,17 +89,20 @@ interface Challenge {
 
 ### Supported Models
 
-- **OpenAI**: `gpt-5`, `gpt-5-mini`, `qwq-32b`, `qwen3-30b`, `qwen3-coder-480b`
-- **Google**: `gemini-2.5-flash`, `gemini-2.5-pro`
-- **xAI**: `grok-4`, `grok-4-0709`
-- **Perplexity**: `sonar-pro`, `perplexity-sonar-pro`
+| Provider | Models | Notes |
+|----------|--------|-------|
+| **Google Gemini** | `gemini-3-pro-preview`, `gemini-2.5-pro`, `gemini-2.5-flash`, `gemini-2.5-flash-lite` | Gemini 3 Pro is latest (Nov 2025) |
+| **OpenAI** | `gpt-5.1`, `gpt-5.1-codex-mini`, `gpt-5.1-codex`, `gpt-5-pro` | Codex models use /v1/responses endpoint |
+| **xAI (Grok)** | `grok-4-1-fast-reasoning`, `grok-4-1-fast-non-reasoning`, `grok-code-fast-1`, `grok-4-0709` | Grok 4.1 is latest (Nov 2025) |
+| **Perplexity** | `sonar-pro`, `sonar-reasoning-pro` | Web search enabled |
+| **OpenRouter** | `qwen/qwen3-coder-plus`, `moonshotai/kimi-k2-thinking` | Requires OPENROUTER_API_KEY |
 
 ### Notes
 
 - Higher `temperature` values produce more diverse and creative challenges
 - The tool automatically detects claim types (fact, opinion, assumption, conclusion)
 - Groupthink detection works best with array contexts containing multiple similar statements
-- Default model (`gpt-5-mini`) balances cost and quality
+- Default model (`gpt-5.1-codex-mini`) balances cost and quality
 
 ---
 
@@ -123,32 +126,32 @@ The Verifier tool provides multi-model parallel verification with consensus anal
 Each variant uses different models and settings optimized for specific use cases:
 
 #### `quick_verify` (Default)
-- **Models**: `gpt5_mini`, `gemini-2.5-flash`, `qwen3-30b`
+- **Models**: `qwen/qwen3-coder-plus`, `gemini-3-pro-preview`, `gpt-5.1-codex-mini`
 - **Tokens**: 2000
 - **Timeout**: 10000ms
 - **Use case**: Fast verification of simple statements
 
 #### `deep_verify`
-- **Models**: `gpt5`, `qwq-32b`, `gpt5_reason`, `gemini-2.5-pro`, `qwen3-coder-480b`
+- **Models**: `qwen/qwen3-coder-plus`, `gemini-3-pro-preview`, `gpt-5.1`
 - **Tokens**: 6000
 - **Timeout**: 30000ms
 - **Use case**: Complex reasoning and analysis
 
 #### `fact_check`
-- **Models**: `perplexity-sonar-pro`, `gpt5`, `gemini-2.5-pro`
+- **Models**: `qwen/qwen3-coder-plus`, `gemini-3-pro-preview`, `gpt-5.1-codex-mini`
 - **Tokens**: 3000
 - **Timeout**: 15000ms
 - **Include Sources**: Yes (default)
 - **Use case**: Factual verification with citations
 
 #### `code_verify`
-- **Models**: `qwen3-coder-480b`, `gpt5`, `gemini-2.5-pro`
+- **Models**: `qwen/qwen3-coder-plus`, `gemini-3-pro-preview`, `gpt-5.1-codex-mini`
 - **Tokens**: 4000
 - **Timeout**: 20000ms
 - **Use case**: Code correctness verification
 
 #### `security_verify`
-- **Models**: `gpt5`, `qwen3-coder-480b`, `grok-4`
+- **Models**: `qwen/qwen3-coder-plus`, `gemini-3-pro-preview`, `gpt-5.1-codex-mini`
 - **Tokens**: 4000
 - **Timeout**: 20000ms
 - **Use case**: Security vulnerability detection
@@ -175,7 +178,7 @@ const result2 = await verifier.verify(
 const result3 = await verifier.verify(
   'Is this code safe?',
   {
-    model: ['gpt-5', 'gemini-2.5-pro'],
+    model: ['gpt-5.1', 'gemini-2.5-pro'],
     maxTokens: 3000
   }
 );
@@ -248,7 +251,7 @@ The Scout tool provides conditional hybrid intelligence gathering, using Perplex
 #### `research_scout` (Default)
 - **Flow**: `perplexity-first-always`
 - **Perplexity Timeout**: 500ms
-- **Parallel Models**: `gemini-2.5-flash`, `gpt-5-mini`
+- **Parallel Models**: `gemini-3-pro-preview`, `gpt-5.1-codex-mini`
 - **Tokens**: 2500
 - **Max Sources**: 100
 - **Use case**: Comprehensive research with current facts
@@ -271,7 +274,7 @@ The Scout tool provides conditional hybrid intelligence gathering, using Perplex
 #### `quick_scout`
 - **Flow**: `conditional-hybrid`
 - **Perplexity Timeout**: 250ms
-- **Parallel Models**: `gemini-2.5-flash`
+- **Parallel Models**: `gemini-3-pro-preview`, `gpt-5.1-codex-mini`
 - **Tokens**: 1000
 - **Max Sources**: 50
 - **Use case**: Fast information gathering
@@ -451,7 +454,7 @@ See test files for more usage examples:
 1. **Use appropriate variants**: Don't use `deep_verify` when `quick_verify` suffices
 2. **Set token limits**: Lower `maxTokens` for simple queries
 3. **Control timeouts**: Shorter timeouts for time-sensitive operations
-4. **Choose models wisely**: `gpt-5-mini` and `gemini-2.5-flash` are fast and cheap
+4. **Choose models wisely**: `gpt-5.1-codex-mini` and `gemini-2.5-flash` are fast and cheap
 5. **Limit Grok sources**: Keep `maxSearchSources` low unless needed
 6. **Use `quick_scout`**: For simple lookups instead of full research
 
@@ -467,7 +470,7 @@ If migrating from old tool structure:
 await thinkTool.challenge(context);
 
 // New
-await challenger.challenge(context, { model: 'gpt-5-mini' });
+await challenger.challenge(context, { model: 'gpt-5.1-codex-mini' });
 ```
 
 ### Verifier (formerly consensus tools)

package/docs/WORKFLOWS.md

@@ -561,7 +561,7 @@ steps:
           problem: "GraphQL architecture patterns"
           style: "systematic"
           quantity: 5
-          model: "gpt-5-mini"
+          model: "gpt-5.1-codex-mini"
         output:
           variable: openai_patterns
 
@@ -587,7 +587,7 @@ steps:
       problem: "Design a distributed caching system"
       style: "innovative"
       quantity: 3
-      model: "gpt-5-mini"
+      model: "gpt-5.1-codex-mini"
     output: initial_design
 
   # Step 2: Challenge the design
@@ -603,7 +603,7 @@ steps:
       query: "Improve the design considering these challenges: ${challenges}"
       mode: "architecture-debate"
       domain: "backend"
-      models: ["gpt-5-mini", "gemini-2.5-pro", "grok-4"]
+      models: ["gpt-5.1-codex-mini", "gemini-2.5-pro", "grok-4"]
       rounds: 5
       pingPongStyle: "debate"
       temperature: 0.7
@@ -614,7 +614,7 @@ steps:
     params:
       query: "Verify this refined design addresses the challenges: ${refined_design}"
       variant: "code_verify"
-      models: ["gpt-5-mini", "gemini-2.5-pro"]
+      models: ["gpt-5.1-codex-mini", "gemini-2.5-pro"]
     output: final_verdict
 ```
 
@@ -690,7 +690,7 @@ steps:
       problem: "${query}"
       style: "systematic"
       quantity: 3
-      model: "gpt-5-mini"
+      model: "gpt-5.1-codex-mini"
     output: gpt_perspective
 
   - tool: perplexity_ask
@@ -723,7 +723,7 @@ steps:
     params:
       query: "Based on synchronized analysis (${sync_analysis}) and latest data (${latest_data}), provide comprehensive answer to: ${query}"
       mode: "deep-reasoning"
-      models: ["gpt-5-mini", "gemini-2.5-pro", "perplexity"]
+      models: ["gpt-5.1-codex-mini", "gemini-2.5-pro", "perplexity"]
       rounds: 5
       pingPongStyle: "collaborative"
     output: deep_analysis
@@ -733,7 +733,7 @@ steps:
     params:
       query: "Challenge and improve this analysis: ${deep_analysis}"
       mode: "architecture-debate"
-      models: ["gpt-5-mini", "gemini-2.5-pro", "grok-4"]
+      models: ["gpt-5.1-codex-mini", "gemini-2.5-pro", "grok-4"]
       rounds: 3
       pingPongStyle: "debate"
     output: challenged_analysis

package/package.json

@@ -2,7 +2,7 @@
   "name": "tachibot-mcp",
   "mcpName": "io.github.byPawel/tachibot-mcp",
   "displayName": "TachiBot MCP - Universal AI Orchestrator",
-  "version": "2.0.6",
+  "version": "2.1.0",
   "type": "module",
   "main": "dist/src/server.js",
   "bin": {

package/profiles/balanced.json

@@ -13,8 +13,7 @@
     "grok_architect": false,
     "grok_brainstorm": false,
     "grok_search": true,
-    "openai_gpt5_reason": true,
-    "openai_compare": false,
+    "openai_reason": true,
     "openai_brainstorm": true,
     "openai_code_review": false,
     "openai_explain": false,

package/profiles/code_focus.json

@@ -13,8 +13,7 @@
     "grok_architect": false,
     "grok_brainstorm": false,
     "grok_search": false,
-    "openai_gpt5_reason": false,
-    "openai_compare": false,
+    "openai_reason": false,
     "openai_brainstorm": false,
     "openai_code_review": true,
     "openai_explain": false,

package/profiles/debug_intensive.json

@@ -22,7 +22,6 @@
     "grok_architect": false,
     "grok_brainstorm": false,
     "grok_search": false,
-    "openai_compare": false,
     "openai_brainstorm": false,
     "gemini_brainstorm": false,
     "gemini_analyze_text": false,

package/profiles/full.json

@@ -1,5 +1,5 @@
 {
-  "description": "All tools enabled for maximum capability (~Xk tokens, 32 tools)",
+  "description": "All tools enabled for maximum capability (~Xk tokens, 31 tools)",
   "tools": {
     "think": true,
     "focus": true,
@@ -13,8 +13,7 @@
     "grok_architect": true,
     "grok_brainstorm": true,
     "grok_search": true,
-    "openai_gpt5_reason": true,
-    "openai_compare": true,
+    "openai_reason": true,
     "openai_brainstorm": true,
     "openai_code_review": true,
     "openai_explain": true,

package/profiles/minimal.json

@@ -13,8 +13,7 @@
     "grok_architect": false,
     "grok_brainstorm": false,
     "grok_search": false,
-    "openai_gpt5_reason": false,
-    "openai_compare": false,
+    "openai_reason": false,
     "openai_brainstorm": false,
     "openai_code_review": false,
     "openai_explain": false,

package/profiles/research_power.json

@@ -13,8 +13,7 @@
     "grok_architect": false,
     "grok_brainstorm": false,
     "grok_search": true,
-    "openai_gpt5_reason": false,
-    "openai_compare": false,
+    "openai_reason": false,
     "openai_brainstorm": true,
     "openai_code_review": false,
     "openai_explain": false,

package/profiles/workflow_builder.json

@@ -13,8 +13,7 @@
     "grok_architect": false,
     "grok_brainstorm": false,
     "grok_search": false,
-    "openai_gpt5_reason": false,
-    "openai_compare": false,
+    "openai_reason": false,
     "openai_brainstorm": false,
     "openai_code_review": false,
     "openai_explain": false,

package/tools.config.json

@@ -25,7 +25,7 @@
       "grok_brainstorm",
       "grok_search"
     ],
-    "openai": ["openai_compare", "openai_brainstorm"],
+    "openai": ["openai_brainstorm", "openai_reason", "openai_code_review", "openai_explain"],
     "gemini": [
       "gemini_brainstorm",
       "gemini_analyze_code",
@@ -38,7 +38,12 @@
       "workflow",
       "list_workflows",
       "create_workflow",
-      "visualize_workflow"
+      "visualize_workflow",
+      "workflow_start",
+      "continue_workflow",
+      "workflow_status",
+      "validate_workflow",
+      "validate_workflow_file"
     ],
     "collaborative": ["pingpong", "qwen_competitive"]
   },
@@ -59,8 +64,10 @@
       "grok_architect": true,
       "grok_brainstorm": true,
       "grok_search": true,
-      "openai_compare": true,
       "openai_brainstorm": true,
+      "openai_reason": true,
+      "openai_code_review": true,
+      "openai_explain": true,
       "gemini_brainstorm": true,
       "gemini_analyze_code": true,
       "gemini_analyze_text": true,
@@ -74,6 +81,11 @@
       "list_workflows": true,
       "create_workflow": true,
       "visualize_workflow": true,
+      "workflow_start": true,
+      "continue_workflow": true,
+      "workflow_status": true,
+      "validate_workflow": true,
+      "validate_workflow_file": true,
       "pingpong": true,
       "qwen_competitive": false
     }

package/workflows/code-architecture-review.yaml

@@ -176,10 +176,12 @@ steps:
   # ═══════════════════════════════════════════════════════════════════════════
 
   - name: consensus
-    tool: openai_compare
+    tool: openai_brainstorm
     input:
-      topic: "Final architecture recommendations for: ${query}"
-      options:
+      problem: |
+        Synthesize final architecture recommendations for: ${query}
+
+        Combine these expert analyses into actionable recommendations:
         - "Grok's SOLID analysis: ${grok-solid-analysis.output}"
         - "Gemini's pattern analysis: ${gemini-pattern-analysis.output}"
         - "Qwen's CQRS evaluation: ${qwen-cqrs-evaluation.output}"

package/workflows/creative-brainstorm-yaml.yaml

@@ -15,7 +15,7 @@ variables:
 steps:
   # Step 1: Claude Thinking - Problem Framing
   - name: claude-thinking
-    tool: openai_gpt5_reason  # Using GPT-5 Mini for structured thinking
+    tool: openai_reason  # Using GPT-5 Mini for structured thinking
     input:
       query: |
         Analyze and structure the brainstorming request: ${query}