minovative-mind-cli 2.8.0 → 2.8.1

package/dist/services/agent.js

@@ -26,7 +26,7 @@ import { markedTerminal } from 'marked-terminal';
 const execAsync = promisify(exec);
 import { debugLog, isDebugOn } from '../utils/logger.js';
 import { ensureProjectStorage, ensureIgnored, readCache, writeCache, invalidateCacheForDependents, } from '../utils/projectStorage.js';
-import { GEMINI_MODELS, isByokEnabled } from '../utils/config.js';
+import { GEMINI_MODELS, isByokEnabled, TPM_COOLING_DELAYS } from '../utils/config.js';
 import { createSharedChatSession, getGeneralChatConfig, getPlanExecutionConfig, compressTextUsingFlashLite, generateChatTitle, getGlobalActiveModel, setGlobalActiveModel, summarizeChatHistory, } from './ai.js';
 import { getAndResetTurnUsage } from './proxyClient.js';
 import { changeLogger } from './changeLogger.js';
@@ -317,6 +317,8 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
             inputHandler.start(spinner);
             changeLogger.startChangeSet(userInput);
             let finalInput = userInput;
+            // Inter-turn cooling-off delay to allow API token buckets to settle before running intent routing
+            await new Promise((resolve) => setTimeout(resolve, TPM_COOLING_DELAYS.INTER_TURN_MS));
             // Stage 0: Summarize chat history if length threshold is met
             if (!cachedContextResult) {
                 const isDebug = isDebugOn();
@@ -359,6 +361,8 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
                     toolLogs.forEach((log) => p.log.step(log));
                     spinner.start(`🔍 Context gathered successfully.`);
                 }
+                // Post-investigation cooling-off delay to allow API Tokens-Per-Minute (TPM) sliding window to settle
+                await new Promise((resolve) => setTimeout(resolve, TPM_COOLING_DELAYS.POST_INVESTIGATION_MS));
             }
             let latestUsage = undefined;
             // Collect any inputs that were queued while the Context Agent was investigating
@@ -481,7 +485,8 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
                             modifiedFiles.forEach((file) => p.log.message(`  ${pc.dim('•')} ${file}`));
                         }
                         else {
-                            summaryMsg = 'Generation was stopped by the user before completion. No code files were modified before stopping.';
+                            summaryMsg =
+                                'Generation was stopped by the user before completion. No code files were modified before stopping.';
                             p.log.warn(pc.yellow('\n⚠️  Generation stopped by user before completion (no files were modified).'));
                         }
                         chat.addTurn({
@@ -657,8 +662,7 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
             const turnDuration = ((turnEndTime - turnStartTime) / 1000).toFixed(1);
             p.log.info(`${pc.dim('Generated in')} ${pc.cyan(turnDuration + 's')}`);
             // Check if generation was stopped by user
-            const isStopped = finalText.includes('Generation stopped') ||
-                finalText.includes('[Generation stopped');
+            const isStopped = finalText.includes('Generation stopped') || finalText.includes('[Generation stopped');
             if (isStopped) {
                 const currentChanges = changeLogger.getCurrentChangeSet()?.changes || [];
                 const modifiedFiles = currentChanges
@@ -672,7 +676,8 @@ export async function executeSingleTurn(workspaceRoot, userInput, chat, inputHan
                     modifiedFiles.forEach((file) => p.log.message(`  ${pc.dim('•')} ${file}`));
                 }
                 else {
-                    summaryMsg = 'Generation was stopped by the user before completion. No code files were modified before stopping.';
+                    summaryMsg =
+                        'Generation was stopped by the user before completion. No code files were modified before stopping.';
                     p.log.warn(pc.yellow('\n⚠️  Generation stopped by user before completion (no files were modified).'));
                 }
                 chat.addTurn({

package/dist/services/orchestration/investigationOrchestrator.js

@@ -25,6 +25,7 @@ import { InvestigationAgentRunner } from './investigationAgent.js';
 import { readFile } from '../agent-tools.js';
 import { buildDependencyGraph } from '../../utils/dependencyTracer.js';
 import { debugLog } from '../../utils/logger.js';
+import { TPM_COOLING_DELAYS } from '../../utils/config.js';
 // ─── Constants ───────────────────────────────────────────────────────
 /** Maximum total relevant files across all agents after merge. */
 const MAX_TOTAL_FILES = 30;
@@ -57,6 +58,8 @@ export class InvestigationOrchestrator {
         const agents = agentAssignments.map((assignment) => new InvestigationAgentRunner(assignment.agentLabel, assignment.domains, workspaceRoot, readCache, projectTree, projectType));
         // 3. Run all agents in parallel
         const startTime = Date.now();
+        // Cooling-off pause to prevent burst 429 errors when initiating parallel agent dispatch
+        await new Promise((resolve) => setTimeout(resolve, TPM_COOLING_DELAYS.PARALLEL_DISPATCH_MS));
         // Limit parallel investigation agents to 2 to avoid Vertex AI RESOURCE_EXHAUSTED errors
         const MAX_CONCURRENT = 2;
         const results = [];

package/dist/services/proxyClient.js

@@ -41,7 +41,7 @@ let globalSessionAccumulatedUsage = {
     totalTokenCount: 0,
     creditsUsed: 0,
     remainingBalance: undefined,
-    modelsUsed: {}
+    modelsUsed: {},
 };
 export function getAndResetTurnUsage() {
     const current = { ...globalSessionAccumulatedUsage };
@@ -52,7 +52,7 @@ export function getAndResetTurnUsage() {
         totalTokenCount: 0,
         creditsUsed: 0,
         remainingBalance: undefined,
-        modelsUsed: {}
+        modelsUsed: {},
     };
     return current;
 }
@@ -117,6 +117,7 @@ export class ProxyClient {
             if ((response.status === 429 || response.status === 503) && attempt < MAX_RETRIES) {
                 const exponentialDelay = Math.min(MAX_DELAY_MS, BASE_DELAY_MS * Math.pow(2, attempt));
                 const delayTime = Math.round(exponentialDelay * (0.5 + Math.random() * 0.5));
+                process.stdout.write('\n');
                 console.warn(`Rate limit or service unavailable hit (${response.status}). Retrying in ${(delayTime / 1000).toFixed(1)}s... (Attempt ${attempt + 1}/${MAX_RETRIES})`);
                 await delay(delayTime, abortSignal);
                 attempt++;
@@ -198,7 +199,7 @@ export class ProxyClient {
                                     collector?.accumulateUsage({
                                         promptTokens: data.usage.promptTokens || 0,
                                         candidatesTokens: data.usage.candidatesTokens || 0,
-                                        cachedTokens: data.usage.cachedTokens || 0
+                                        cachedTokens: data.usage.cachedTokens || 0,
                                     });
                                     globalSessionAccumulatedUsage.promptTokens += data.usage.promptTokens || 0;
                                     globalSessionAccumulatedUsage.candidatesTokens += data.usage.candidatesTokens || 0;
@@ -209,7 +210,8 @@ export class ProxyClient {
                                     if (data.usage.remainingBalance !== undefined) {
                                         globalSessionAccumulatedUsage.remainingBalance = data.usage.remainingBalance;
                                     }
-                                    globalSessionAccumulatedUsage.modelsUsed[modelName] = (globalSessionAccumulatedUsage.modelsUsed[modelName] || 0) + 1;
+                                    globalSessionAccumulatedUsage.modelsUsed[modelName] =
+                                        (globalSessionAccumulatedUsage.modelsUsed[modelName] || 0) + 1;
                                 }
                                 if (data.groundingMetadata) {
                                     groundingMetadata = data.groundingMetadata;
@@ -236,6 +238,7 @@ export class ProxyClient {
                     if (attempt < MAX_RETRIES) {
                         const exponentialDelay = Math.min(MAX_DELAY_MS, BASE_DELAY_MS * Math.pow(2, attempt));
                         const delayTime = Math.round(exponentialDelay * (0.5 + Math.random() * 0.5));
+                        process.stdout.write('\n');
                         console.warn(`Rate limit or service unavailable hit during stream. Retrying in ${(delayTime / 1000).toFixed(1)}s... (Attempt ${attempt + 1}/${MAX_RETRIES})`);
                         await delay(delayTime, abortSignal);
                         attempt++;

package/dist/utils/config.d.ts

@@ -25,6 +25,19 @@ export declare const GEMINI_MODELS: {
 export declare const DEFAULT_MODEL: "auto";
 /** Maximum tokens the model can output per response. */
 export declare const MAX_OUTPUT_TOKENS = 60000;
+/**
+ * Rate limit & Tokens-Per-Minute (TPM) cooling-off delays (in milliseconds).
+ * Prevents transient 429 / 503 rate-limit errors by giving the API 60-second
+ * rolling token window time to settle during key agent transitions.
+ */
+export declare const TPM_COOLING_DELAYS: {
+    /** Pause before running initial intent routing between turns */
+    readonly INTER_TURN_MS: 1000;
+    /** Pause before dispatching parallel sub-agent investigations simultaneously */
+    readonly PARALLEL_DISPATCH_MS: 1000;
+    /** Pause after context gathering completes before starting the execution stream */
+    readonly POST_INVESTIGATION_MS: 3000;
+};
 /**
  * Checks if BYOK is currently enabled for the user.
  */

package/dist/utils/config.js

@@ -25,6 +25,19 @@ export const GEMINI_MODELS = {
 export const DEFAULT_MODEL = GEMINI_MODELS.AUTO;
 /** Maximum tokens the model can output per response. */
 export const MAX_OUTPUT_TOKENS = 60_000;
+/**
+ * Rate limit & Tokens-Per-Minute (TPM) cooling-off delays (in milliseconds).
+ * Prevents transient 429 / 503 rate-limit errors by giving the API 60-second
+ * rolling token window time to settle during key agent transitions.
+ */
+export const TPM_COOLING_DELAYS = {
+    /** Pause before running initial intent routing between turns */
+    INTER_TURN_MS: 1000,
+    /** Pause before dispatching parallel sub-agent investigations simultaneously */
+    PARALLEL_DISPATCH_MS: 1000,
+    /** Pause after context gathering completes before starting the execution stream */
+    POST_INVESTIGATION_MS: 3000,
+};
 /**
  * Checks if BYOK is currently enabled for the user.
  */

package/oclif.manifest.json

@@ -65,5 +65,5 @@
       ]
     }
   },
-  "version": "2.8.0"
+  "version": "2.8.1"
 }

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": "2.8.0",
+  "version": "2.8.1",
   "author": "Daniel Ward",
   "bin": {
     "minovative-mind-cli": "bin/run.js"
@@ -78,9 +78,8 @@
   "scripts": {
     "build": "shx rm -rf dist tsconfig.tsbuildinfo && tsc -b",
     "prepare": "npm run build",
-    "lint": "eslint",
+    "lint": "eslint src/ test/",
     "postpack": "shx rm -f oclif.manifest.json",
-    "posttest": "npm run lint",
     "prepack": "npm run build && oclif manifest && oclif readme --no-source-links",
     "test": "mocha --forbid-only \"test/**/*.test.ts\"",
     "version": "oclif readme --no-source-links && git add README.md"