minovative-mind-cli 2.2.4 → 2.2.5

package/README.md

@@ -86,13 +86,12 @@ Works in any terminal — SSH, Docker, CI pipelines, Vim, Windows Command Prompt
 
 Hot-swap during a session using `/models`:
 
-| Model                     | Best for                                            |
-| ------------------------- | --------------------------------------------------- |
+| Model                     | Best for                                       |
+| ------------------------- | ---------------------------------------------- |
 | **Auto** (default)        | Automatically selects (3.5 Flash or 2.5 Flash) |
-| **Gemini 3.5 Flash**      | Everyday coding — fast and accurate                 |
-| **Gemini 2.5 Flash**      | Cost-efficient reasoning and multimodal tasks       |
-| **Gemini 3.1 Pro**        | Complex architectural changes                       |
-| **Gemini 3.1 Flash Lite** | Maximum speed and cost efficiency                   |
+| **Gemini 3.5 Flash**      | Everyday coding — fast and accurate            |
+| **Gemini 3.1 Pro**        | Complex architectural changes                  |
+| **Gemini 3.1 Flash Lite** | Maximum speed and cost efficiency              |
 
 > Background tasks (routing, compression, commits) always use lightweight
 > models automatically. You pay for those lightweight model
@@ -104,21 +103,21 @@ Hot-swap during a session using `/models`:
 
 ## Session Commands
 
-| Command         | What it does                                                     |
-| --------------- | ---------------------------------------------------------------- |
-| `/models`       | Hot-swap the active model                                        |
-| `/plan`         | Toggle plan mode to review implementation strategies             |
-| `/paste`        | Multi-line input mode (cancel with Ctrl+C)                       |
-| `/clear`        | Clear conversation history                                       |
-| `/debug`        | Expose internal agent diagnostics                                |
-| `/auto-approve` | Toggle skipping confirmation prompts for commands                |
-| `/sub-agents`   | Toggle the MMAAK Engine for parallel investigation and execution (concurrency limited to 2) |
-| `/stats`        | View current session statistics and configuration                |
-| `/commit`       | Generate a conventional commit message from your diff            |
-| `/revert`       | Undo changes from the last turn, or toggle the revert logger     |
+| Command         | What it does                                                                                                               |
+| --------------- | -------------------------------------------------------------------------------------------------------------------------- |
+| `/models`       | Hot-swap the active model                                                                                                  |
+| `/plan`         | Toggle plan mode to review implementation strategies                                                                       |
+| `/paste`        | Multi-line input mode (cancel with Ctrl+C)                                                                                 |
+| `/clear`        | Clear conversation history                                                                                                 |
+| `/debug`        | Expose internal agent diagnostics                                                                                          |
+| `/auto-approve` | Toggle skipping confirmation prompts for commands                                                                          |
+| `/sub-agents`   | Toggle the MMAAK Engine for parallel investigation and execution (concurrency limited to 2)                                |
+| `/stats`        | View current session statistics and configuration                                                                          |
+| `/commit`       | Generate a conventional commit message from your diff                                                                      |
+| `/revert`       | Undo changes from the last turn, or toggle the revert logger                                                               |
 | `/chats`        | View, resume, or delete previous chat sessions (with token tracking, Git branch, auto-approve status, and sub-agent state) |
-| `/workspaces`   | Manage active workspaces and linked cross-repo aliases           |
-| `stop`          | Abort generation immediately                                     |
+| `/workspaces`   | Manage active workspaces and linked cross-repo aliases                                                                     |
+| `stop`          | Abort generation immediately                                                                                               |
 
 ---
 

package/dist/services/agent/slashCommands.js

@@ -78,14 +78,14 @@ export async function handleSlashCommand(command, context) {
                     hint: 'Balanced performance',
                 },
                 {
-                    value: 'gemini-2.5-flash',
-                    label: 'Gemini 2.5 Flash',
-                    hint: 'Google Gemini 2.5 Flash model with built-in reasoning',
+                    value: 'gemini-3.1-flash-lite',
+                    label: 'Gemini 3.1 Flash-Lite',
+                    hint: 'Ultra-fast and cost-effective',
                 },
                 {
                     value: 'auto',
-                    label: 'Auto (Gemini 2.5 / Gemini 3.5)',
-                    hint: 'Dynamically routes between Gemini 2.5 and Gemini 3.5 based on prompt complexity',
+                    label: 'Auto (Flash-Lite / Flash)',
+                    hint: 'Dynamically routes between Gemini 3.1 Flash-Lite and Gemini 3.5 based on prompt complexity',
                 },
             ],
         });

package/dist/services/agent.js

@@ -313,7 +313,7 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
             if (getGlobalActiveModel() === GEMINI_MODELS.AUTO) {
                 let selectedModel = GEMINI_MODELS.FLASH_3_5;
                 if (effectiveTargetAgent === 'CHAT') {
-                    selectedModel = GEMINI_MODELS.FLASH_2_5;
+                    selectedModel = GEMINI_MODELS.FLASH_LITE_3_1;
                 }
                 else {
                     spinner.stop(); // MUST clear the investigation spinner first to prevent leaking the setInterval
@@ -321,7 +321,7 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
                     const complexity = await evaluateExecutionComplexity(finalInput, gatherRes.contextResult?.summary, gatherRes.contextResult?.relevantFiles?.size || 0);
                     spinner.stop();
                     process.stdout.write('\x1b[2K\r');
-                    selectedModel = complexity === 'EASY' ? GEMINI_MODELS.FLASH_2_5 : GEMINI_MODELS.FLASH_3_5;
+                    selectedModel = complexity === 'EASY' ? GEMINI_MODELS.FLASH_LITE_3_1 : GEMINI_MODELS.FLASH_3_5;
                 }
                 chat.setModel(selectedModel);
                 debugLog(`Auto-routing to model ${selectedModel} based on intent ${effectiveTargetAgent}`);
@@ -481,6 +481,10 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
                     const percentSaved = totalInputTokens > 0 ? Math.round((latestUsage.cachedTokens / totalInputTokens) * 100) : 0;
                     p.log.info(`${pc.green('⚡')} ${pc.green('Context Cache Hit:')} ${pc.bold(latestUsage.cachedTokens.toLocaleString())} tokens cached ${pc.dim(`(Saved ~${percentSaved}% of input cost)`)}`);
                 }
+                const inputTokens = (latestUsage.promptTokens || 0) + (latestUsage.cachedTokens || 0);
+                const outputTokens = latestUsage.candidatesTokens || 0;
+                const totalTokens = latestUsage.totalTokenCount || inputTokens + outputTokens;
+                p.log.info(`${pc.dim('Tokens Used:')} ${pc.cyan(totalTokens.toLocaleString())} ${pc.dim(`(Input: ${inputTokens.toLocaleString()}, Output: ${outputTokens.toLocaleString()})`)}`);
                 if (latestUsage.remainingBalance !== undefined) {
                     p.log.info(`${pc.dim('Credits Remaining:')} ${pc.cyan(latestUsage.remainingBalance.toLocaleString())}`);
                 }
@@ -494,7 +498,8 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
             }
             let turnTokens = 0;
             if (latestUsage) {
-                turnTokens = (latestUsage.promptTokens || 0) + (latestUsage.cachedTokens || 0) + (latestUsage.candidatesTokens || 0);
+                turnTokens =
+                    (latestUsage.promptTokens || 0) + (latestUsage.cachedTokens || 0) + (latestUsage.candidatesTokens || 0);
             }
             chatSessionState.totalTokens += turnTokens;
             let gitBranch = '';
@@ -583,11 +588,6 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
                                     label: 'Gemini 3.5 Flash',
                                     hint: 'Balanced performance',
                                 },
-                                {
-                                    value: 'gemini-2.5-flash',
-                                    label: 'Gemini 2.5 Flash',
-                                    hint: 'Google Gemini 2.5 Flash model with built-in reasoning',
-                                },
                                 {
                                     value: 'gemini-3.1-flash-lite',
                                     label: 'Gemini 3.1 Flash-Lite',
@@ -595,8 +595,8 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
                                 },
                                 {
                                     value: 'auto',
-                                    label: 'Auto (Gemini 2.5 / Gemini 3.5)',
-                                    hint: 'Dynamically routes between Gemini 2.5 and Gemini 3.5 based on prompt complexity',
+                                    label: 'Auto (Flash-Lite / Flash)',
+                                    hint: 'Dynamically routes between Gemini 3.1 Flash-Lite and Gemini 3.5 based on prompt complexity',
                                 },
                             ],
                         });

package/dist/services/ai.js

@@ -322,7 +322,7 @@ export async function compressTextUsingFlashLite(text, instruction = 'Summarize
             return text;
         let model = getGlobalActiveModel();
         if (model === 'auto')
-            model = GEMINI_MODELS.FLASH_2_5;
+            model = GEMINI_MODELS.FLASH_LITE_3_1;
         const parts = [{ text }];
         if (inlineData) {
             parts.push({ inlineData });
@@ -531,7 +531,7 @@ export function createContextAgentSession() {
 export function createIntentRouterSession() {
     let model = getGlobalActiveModel();
     if (model === 'auto')
-        model = GEMINI_MODELS.FLASH_2_5;
+        model = GEMINI_MODELS.FLASH_LITE_3_1;
     return new ProxyChatSession(model, INTENT_ROUTER_SYSTEM_INSTRUCTION, [], // no tools
     { temperature: 0, responseMimeType: 'application/json' });
 }
@@ -539,7 +539,7 @@ export function createIntentRouterSession() {
 export function createExecutionComplexitySession() {
     let model = getGlobalActiveModel();
     if (model === 'auto')
-        model = GEMINI_MODELS.FLASH_2_5;
+        model = GEMINI_MODELS.FLASH_LITE_3_1;
     return new ProxyChatSession(model, EXECUTION_COMPLEXITY_SYSTEM_INSTRUCTION, [], // no tools
     { temperature: 0, responseMimeType: 'application/json' });
 }
@@ -547,7 +547,7 @@ export function createExecutionComplexitySession() {
 export function createInvestigationComplexitySession() {
     let model = getGlobalActiveModel();
     if (model === 'auto')
-        model = GEMINI_MODELS.FLASH_2_5;
+        model = GEMINI_MODELS.FLASH_LITE_3_1;
     return new ProxyChatSession(model, INVESTIGATION_COMPLEXITY_SYSTEM_INSTRUCTION, [], // no tools
     { temperature: 0, responseMimeType: 'application/json' });
 }

package/dist/services/investigationComplexity.d.ts

@@ -31,7 +31,7 @@ export interface InvestigationComplexityResult {
 /**
  * Evaluates whether the investigation phase should be parallelized.
  *
- * Uses `gemini-2.5-flash` (under auto mode) at temperature 0 to classify the prompt's
+ * Uses `gemini-3.1-flash-lite` (under auto mode) at temperature 0 to classify the prompt's
  * investigation complexity. The PM dynamically identifies domains and groups
  * them into agent assignments. Agent count = `agentAssignments.length`, which
  * may be fewer than `domains.length` when related domains are batched together.

package/dist/services/investigationComplexity.js

@@ -16,7 +16,7 @@ import { debugLog } from '../utils/logger.js';
 /**
  * Evaluates whether the investigation phase should be parallelized.
  *
- * Uses `gemini-2.5-flash` (under auto mode) at temperature 0 to classify the prompt's
+ * Uses `gemini-3.1-flash-lite` (under auto mode) at temperature 0 to classify the prompt's
  * investigation complexity. The PM dynamically identifies domains and groups
  * them into agent assignments. Agent count = `agentAssignments.length`, which
  * may be fewer than `domains.length` when related domains are batched together.

package/dist/services/orchestration/investigationAgent.d.ts

@@ -32,6 +32,8 @@ export interface InvestigationResult {
     webSearchSummary: string;
     /** Total tokens consumed by this agent */
     creditsUsed: number;
+    inputTokens?: number;
+    outputTokens?: number;
     /** Whether the agent completed successfully */
     success: boolean;
 }
@@ -55,6 +57,8 @@ export declare class InvestigationAgentRunner {
     private chat;
     private lastHeartbeat;
     private creditsUsed;
+    private inputTokens;
+    private outputTokens;
     /** Shorter stall timeout for investigation agents (120s to allow for heavy vision/web tasks) */
     static readonly STALL_TIMEOUT_MS = 120000;
     constructor(agentLabel: string, domains: string[], workspaceRoot: string, readCache: ReadCache, projectTree: string, projectType: string);

package/dist/services/orchestration/investigationAgent.js

@@ -38,6 +38,8 @@ export class InvestigationAgentRunner {
     chat;
     lastHeartbeat = Date.now();
     creditsUsed = 0;
+    inputTokens = 0;
+    outputTokens = 0;
     /** Shorter stall timeout for investigation agents (120s to allow for heavy vision/web tasks) */
     static STALL_TIMEOUT_MS = 120_000;
     constructor(agentLabel, domains, workspaceRoot, readCache, projectTree, projectType) {
@@ -322,7 +324,7 @@ export class InvestigationAgentRunner {
             crashed = true;
             summary = 'Investigation aborted by user.';
         }
-        debugLog(`InvestigationAgent [${this.agentLabel}]: Finished. Success=${success && !crashed}, Files=${relevantFiles.size}, Tokens=${this.creditsUsed}`);
+        debugLog(`InvestigationAgent [${this.agentLabel}]: Finished. Success=${success && !crashed}, Files=${relevantFiles.size}, Tokens=${this.creditsUsed} (In: ${this.inputTokens}, Out: ${this.outputTokens})`);
         return {
             domains: this.domains,
             agentLabel: this.agentLabel,
@@ -330,6 +332,8 @@ export class InvestigationAgentRunner {
             summary,
             webSearchSummary,
             creditsUsed: this.creditsUsed,
+            inputTokens: this.inputTokens,
+            outputTokens: this.outputTokens,
             success: success && !crashed,
         };
     }
@@ -339,10 +343,12 @@ export class InvestigationAgentRunner {
     updateUsage(result) {
         const usage = result.response.usageMetadata?.();
         if (usage) {
-            const tokens = usage.totalTokenCount || ((usage.promptTokens || 0) + (usage.candidatesTokens || 0) + (usage.cachedTokens || 0));
-            if (tokens) {
-                this.creditsUsed += tokens;
-            }
+            const input = (usage.promptTokens || 0) + (usage.cachedTokens || 0);
+            const output = usage.candidatesTokens || 0;
+            const tokens = usage.totalTokenCount || (input + output);
+            this.inputTokens += input;
+            this.outputTokens += output;
+            this.creditsUsed += tokens;
         }
     }
 }

package/dist/services/orchestration/investigationOrchestrator.js

@@ -98,12 +98,14 @@ export class InvestigationOrchestrator {
         // 7. Log summary
         const cacheStats = readCache.getStats();
         const totalTokens = results.reduce((sum, r) => sum + r.creditsUsed, 0);
+        const totalInputTokens = results.reduce((sum, r) => sum + (r.inputTokens || 0), 0);
+        const totalOutputTokens = results.reduce((sum, r) => sum + (r.outputTokens || 0), 0);
         const totalFilesBeforeDedup = results.reduce((sum, r) => sum + r.relevantFiles.size, 0);
         p.log.info(`${pc.green('✓')} Parallel Investigation complete.\n` +
             `  Agents: ${agentCount} (${failedResults.length} failed) | ` +
             `Files: ${mergedResult.relevantFiles.size} (deduped from ${totalFilesBeforeDedup}) | ` +
             `Cache hits: ${cacheStats.hitCount}\n` +
-            `  Duration: ${duration}s | Tokens: ${totalTokens.toLocaleString()}`);
+            `  Duration: ${duration}s | Tokens: ${totalTokens.toLocaleString()} ${pc.dim(`(Input: ${totalInputTokens.toLocaleString()}, Output: ${totalOutputTokens.toLocaleString()})`)}`);
         return mergedResult;
     }
     /**

package/dist/services/orchestration/orchestrator.js

@@ -226,10 +226,14 @@ export class Orchestrator {
         p.log.step(pc.cyan('Orchestrator: Reconciling results...'));
         const stats = this.bus.getStats();
         let totalTokens = 0;
+        let inputTokens = 0;
+        let outputTokens = 0;
         let failedTasks = 0;
         let rawSummaryData = '';
         for (const [taskId, res] of this.agentResults.entries()) {
             totalTokens += res.creditsUsed;
+            inputTokens += (res.inputTokens || 0);
+            outputTokens += (res.outputTokens || 0);
             if (!res.success)
                 failedTasks++;
             debugLog(`Task ${taskId} summary: ${res.summary.substring(0, 100)}...`);
@@ -249,7 +253,7 @@ export class Orchestrator {
         p.log.info(`${pc.green('✓')} Sub-agent execution complete.\n` +
             `  Total tasks: ${graph.tasks.length} (${failedTasks} failed)\n` +
             `  Bus Activity: ${stats.activityCount} actions, ${stats.signalCount} signals\n` +
-            `  Total Tokens: ${totalTokens.toLocaleString()}`);
+            `  Total Tokens: ${totalTokens.toLocaleString()} ${pc.dim(`(Input: ${inputTokens.toLocaleString()}, Output: ${outputTokens.toLocaleString()})`)}`);
         // Clear the bus persistence now that orchestration is done
         await this.bus.cleanup();
         this.locks.shutdown();

package/dist/services/orchestration/subAgent.d.ts

@@ -16,6 +16,8 @@ export interface SubAgentResult {
     summary: string;
     /** Total credits (tokens) consumed by this agent */
     creditsUsed: number;
+    inputTokens?: number;
+    outputTokens?: number;
     /** Did the agent hit a stall timeout or crash? */
     crashed: boolean;
 }
@@ -34,6 +36,8 @@ export declare class SubAgentRunner {
     private chat;
     private lastHeartbeat;
     private creditsUsed;
+    private inputTokens;
+    private outputTokens;
     /** Max time without a tool call or response before the agent is considered stalled */
     static readonly STALL_TIMEOUT_MS = 60000;
     constructor(taskId: string, intent: string, workspaceRoot: string, bus: MessageBus, locks: FileLockRegistry, globalContext: string, onProgress?: ((msg: string) => void) | undefined);

package/dist/services/orchestration/subAgent.js

@@ -24,6 +24,8 @@ export class SubAgentRunner {
     chat;
     lastHeartbeat = Date.now();
     creditsUsed = 0;
+    inputTokens = 0;
+    outputTokens = 0;
     /** Max time without a tool call or response before the agent is considered stalled */
     static STALL_TIMEOUT_MS = 60_000;
     constructor(taskId, intent, workspaceRoot, bus, locks, globalContext, onProgress) {
@@ -180,11 +182,13 @@ export class SubAgentRunner {
                 crashed = true;
                 finalSummary = 'Aborted by orchestrator.';
             }
-            debugLog(`SubAgent [${this.taskId}]: Finished. Success=${success}, Crashed=${crashed}, Tokens=${this.creditsUsed}`);
+            debugLog(`SubAgent [${this.taskId}]: Finished. Success=${success}, Crashed=${crashed}, Tokens=${this.creditsUsed} (In: ${this.inputTokens}, Out: ${this.outputTokens})`);
             return {
                 success: success && !crashed,
                 summary: finalSummary,
                 creditsUsed: this.creditsUsed,
+                inputTokens: this.inputTokens,
+                outputTokens: this.outputTokens,
                 crashed,
             };
         });
@@ -195,10 +199,12 @@ export class SubAgentRunner {
     updateUsage(result) {
         const usage = result.response.usageMetadata?.();
         if (usage) {
-            const tokens = usage.totalTokenCount || ((usage.promptTokens || 0) + (usage.candidatesTokens || 0) + (usage.cachedTokens || 0));
-            if (tokens) {
-                this.creditsUsed += tokens;
-            }
+            const input = (usage.promptTokens || 0) + (usage.cachedTokens || 0);
+            const output = usage.candidatesTokens || 0;
+            const tokens = usage.totalTokenCount || (input + output);
+            this.inputTokens += input;
+            this.outputTokens += output;
+            this.creditsUsed += tokens;
         }
     }
 }

package/dist/utils/config.d.ts

@@ -16,7 +16,6 @@ export declare const GITHUB_CLIENT_ID = "Ov23linFYFfjO3JILG7r";
 export declare const GEMINI_MODELS: {
     readonly PRO_3_1: "gemini-3.1-pro-preview";
     readonly FLASH_3_5: "gemini-3.5-flash";
-    readonly FLASH_2_5: "gemini-2.5-flash";
     readonly FLASH_LITE_3_1: "gemini-3.1-flash-lite";
     readonly AUTO: "auto";
 };

package/dist/utils/config.js

@@ -16,7 +16,6 @@ export const GITHUB_CLIENT_ID = 'Ov23linFYFfjO3JILG7r';
 export const GEMINI_MODELS = {
     PRO_3_1: 'gemini-3.1-pro-preview',
     FLASH_3_5: 'gemini-3.5-flash',
-    FLASH_2_5: 'gemini-2.5-flash',
     FLASH_LITE_3_1: 'gemini-3.1-flash-lite',
     AUTO: 'auto',
 };

package/oclif.manifest.json

@@ -65,5 +65,5 @@
       ]
     }
   },
-  "version": "2.2.4"
+  "version": "2.2.5"
 }

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "minovative-mind-cli",
   "description": "An automated AI agent powered by Vertex AI that helps you write software",
-  "version": "2.2.4",
+  "version": "2.2.5",
   "author": "Daniel Ward",
   "bin": {
     "minovative-mind-cli": "bin/run.js"