fix: revert bad ctx trimming add n_keep>n_ctx GPU error detection

.astra/tests/stress/.astra/cache/259a37934ead3910a8722b82054d46d2ca2057b05c488be1dcf439166ac5a9a1.json

@@ -1,5 +1,5 @@
 {
   "result": "Final report with inconsistencies. This should be long enough to pass validation.",
-  "createdAt": 1778129530764,
+  "createdAt": 1778130178728,
   "modelVersion": "unknown"
 }

.astra/tests/stress/.astra/cache/65775be352df43297b63c7af59c9f4f39d2bc368f77456c37b5eef9a94a66b5c.json

@@ -1,5 +1,5 @@
 {
   "result": "[CONFLICT WARNING] 성능이 200% 증가했습니다. vs 그러나 동시에 50% 감소했습니다. 최적화와 성능 저하가 동시에 발견됨.",
-  "createdAt": 1778129530762,
+  "createdAt": 1778130178726,
   "modelVersion": "unknown"
 }

.astra/tests/stress/.astra/cache/6894d26c5b0a55d25d756a473225c7a44d7661af673b24e3f49551a7a2e50280.json

@@ -1,5 +1,5 @@
 {
   "result": "Detailed Execution Plan: 1. Research 2. Analyze 3. Write report with high quality.",
-  "createdAt": 1778129530759,
+  "createdAt": 1778130178723,
   "modelVersion": "unknown"
 }

.astra/tests/stress/.astra/cache/88cb61499f88ed38165b64bd3e8adc543795e4b427b64540a49c9ab27c7fe213.json

@@ -1,5 +1,5 @@
 {
-  "result": "---\nid: stress_conflict_1778129530743\ndate: 2026-05-07T04:52:10.766Z\ntype: knowledge_artifact\nstandard: P-Reinforce v3.0\ntags: [automated, connect_ai, brain_sync]\n---\n\n## 📌 Brief Summary\nFinal report with inconsistencies. This should be long enough to pass validation.\n\nFinal report with inconsistencies. This should be long enough to pass validation.\n\n---\n## 💡 Astra의 선제적 제안 (Proactive Next Actions)\nFinal report with inconsistencies. This should be long enough to pass validation.\n---\n## 🛡️ Reliability & Audit Summary\n> [!NOTE]\n> 이 문서는 ConnectAI의 **Intelligent Resilience** 엔진에 의해 검증 및 정제되었습니다.\n\n| Metric | Value | Status |\n| :--- | :--- | :--- |\n| **Conflict Risk** | `60/100` | ⚠️ Medium |\n| **Fallbacks Used** | `0` | ✅ None |\n| **Auto Retries** | `0` | ✅ Stable |\n| **Deduplication** | `0` | Standard |\n| **Processing Time** | `0.0s` | ✅ Fast |\n\n### 🔍 Decision Audit Trail\n- **[PLANNER]** 전략 수립 중... (15ms)\n- **[RESEARCHER]** 핵심 정보 수집 및 분석 중... (3ms)\n- **[WRITER]** 최종 리포트 작성 및 편집 중... (2ms)\n",
+  "result": "---\nid: stress_conflict_1778130178706\ndate: 2026-05-07T05:02:58.729Z\ntype: knowledge_artifact\nstandard: P-Reinforce v3.0\ntags: [automated, connect_ai, brain_sync]\n---\n\n## 📌 Brief Summary\nFinal report with inconsistencies. This should be long enough to pass validation.\n\nFinal report with inconsistencies. This should be long enough to pass validation.\n\n---\n## 💡 Astra의 선제적 제안 (Proactive Next Actions)\nFinal report with inconsistencies. This should be long enough to pass validation.\n---\n## 🛡️ Reliability & Audit Summary\n> [!NOTE]\n> 이 문서는 ConnectAI의 **Intelligent Resilience** 엔진에 의해 검증 및 정제되었습니다.\n\n| Metric | Value | Status |\n| :--- | :--- | :--- |\n| **Conflict Risk** | `60/100` | ⚠️ Medium |\n| **Fallbacks Used** | `0` | ✅ None |\n| **Auto Retries** | `0` | ✅ Stable |\n| **Deduplication** | `0` | Standard |\n| **Processing Time** | `0.0s` | ✅ Fast |\n\n### 🔍 Decision Audit Trail\n- **[PLANNER]** 전략 수립 중... (16ms)\n- **[RESEARCHER]** 핵심 정보 수집 및 분석 중... (3ms)\n- **[WRITER]** 최종 리포트 작성 및 편집 중... (2ms)\n",
-  "createdAt": 1778129530766,
+  "createdAt": 1778130178730,
   "modelVersion": "unknown"
 }

.astra/tests/stress/.astra/missions/stress_conflict_1778130178706.json

@@ -1,8 +1,8 @@
 {
-  "missionId": "stress_conflict_1778129530743",
+  "missionId": "stress_conflict_1778130178706",
   "status": "completed",
-  "startTime": "2026-05-07T04:52:10.743Z",
+  "startTime": "2026-05-07T05:02:58.706Z",
-  "totalElapsedMs": 24,
+  "totalElapsedMs": 25,
   "results": {
     "planner": "Detailed Execution Plan: 1. Research 2. Analyze 3. Write report with high quality.",
     "researcher": "[CONFLICT WARNING] 성능이 200% 증가했습니다. vs 그러나 동시에 50% 감소했습니다. 최적화와 성능 저하가 동시에 발견됨.",
@@ -16,30 +16,30 @@
     {
       "from": "idle",
       "to": "planner",
-      "durationMs": 15,
+      "durationMs": 16,
       "message": "전략 수립 중...",
-      "ts": "2026-05-07T04:52:10.758Z"
+      "ts": "2026-05-07T05:02:58.722Z"
     },
     {
       "from": "planner",
       "to": "researcher",
       "durationMs": 3,
       "message": "핵심 정보 수집 및 분석 중...",
-      "ts": "2026-05-07T04:52:10.761Z"
+      "ts": "2026-05-07T05:02:58.725Z"
     },
     {
       "from": "researcher",
       "to": "writer",
       "durationMs": 2,
       "message": "최종 리포트 작성 및 편집 중...",
-      "ts": "2026-05-07T04:52:10.763Z"
+      "ts": "2026-05-07T05:02:58.727Z"
     },
     {
       "from": "writer",
       "to": "completed",
       "durationMs": 4,
       "message": "미션 완료",
-      "ts": "2026-05-07T04:52:10.767Z"
+      "ts": "2026-05-07T05:02:58.731Z"
     }
   ],
   "resilienceMetrics": {

package.json

@@ -2,7 +2,7 @@
   "name": "astra",
   "displayName": "Astra",
   "description": "The personal intelligence layer for Antigravity and VS Code. A private cognitive partner for deep project context, memory, and proactive strategic decision-making.",
-  "version": "2.80.9",
+  "version": "2.80.11",
   "publisher": "g1nation",
   "license": "MIT",
   "icon": "assets/icon.png",

src/agent.ts

@@ -526,7 +526,28 @@ export class AgentExecutor {
                     logInfo('Generation aborted by user.');
                 } else {
                     logError('Stream reading error.', { engine, apiUrl, error: err?.message || String(err) });
-                    this.webview?.postMessage({ type: 'error', value: `Connection lost: ${err.message}` });
+                    // LM Studio llama.cpp n_keep > n_ctx 에러 감지
+                    const errMsg = String(err?.message || err || '');
+                    const nCtxMatch = errMsg.match(/n_keep\s*:\s*(\d+)\s*>=?\s*n_ctx\s*:\s*(\d+)/);
+                    if (nCtxMatch) {
+                        const nKeep = nCtxMatch[1];
+                        const nCtx = nCtxMatch[2];
+                        this.webview?.postMessage({
+                            type: 'error',
+                            value: [
+                                `Connection lost: AI Engine Error (LM Studio)`,
+                                `시스템 프롬프트 토큰(${nKeep})이 실제 KV 캐시(${nCtx})보다 큽니다.`,
+                                '',
+                                '**원인:** LM Studio가 GPU 메모리 부족으로 컨텍스트 창을 축소했습니다.',
+                                '**해결 방법 (LM Studio 재설정):**',
+                                `1. LM Studio → 모델 설정 → GPU Offload 레이어를 줄여보세요 (현재 설정보다 낮게)`,
+                                `2. 모델을 언로드 후 재로드하세요`,
+                                `3. 다른 앱이 GPU 메모리를 점유하고 있다면 종료 후 재시도`,
+                            ].join('\n')
+                        });
+                    } else {
+                        this.webview?.postMessage({ type: 'error', value: `Connection lost: ${err.message}` });
+                    }
                 }
             }
 
@@ -2020,40 +2041,26 @@ export class AgentExecutor {
 
             for (const candidateModel of modelCandidates) {
                 for (const variant of messageVariants) {
-                    // LM Studio: context_length를 명시적으로 제한하여 컨텍스트 초과 방지
-                    // 총 메시지 토큰 추정: 문자 수 / 4 (rough estimate)
                     const totalChars = variant.messages.reduce((acc, m) => acc + String(m.content || '').length, 0);
                     const estimatedTokens = Math.ceil(totalChars / 4);
-                    // LM Studio 소형 모델(4B~8B)은 4096~8192 context 제한
-                    // 컨텍스트 초과 시 max_tokens을 줄여서 모델이 응답할 공간 확보
-                    const lmStudioMaxTokens = Math.max(512, Math.min(4096, 8192 - estimatedTokens));
                     const streamBody = {
                         model: candidateModel,
                         messages: variant.messages,
                         stream: true,
                         ...(engine === 'lmstudio'
-                            ? {
+                            ? { max_tokens: 4096, temperature }
-                                max_tokens: lmStudioMaxTokens,
-                                temperature,
-                                // LM Studio: context_length로 컨텍스트 창 명시 설정
-                                context_length: 8192
-                            }
                             : { options: { num_ctx: 32768, num_predict: 4096, temperature } }),
                     };
-                    if (engine === 'lmstudio' && estimatedTokens > 6000) {
+                    logInfo('AI streaming request started.', {
-                        logError('LM Studio context may be too large for small models.', { estimatedTokens, lmStudioMaxTokens, model: candidateModel });
+                        engine, apiUrl, model: candidateModel,
-                    }
+                        variant: variant.name,
+                        messageCount: variant.messages.length,
+                        estimatedTokens,
+                        roles: variant.messages.map(message => message.role),
+                        firstUserPreview: summarizeText(String(variant.messages.find(message => message.role === 'user')?.content || ''), 300)
+                    });
 
                     try {
-                        logInfo('AI streaming request started.', {
-                            engine,
-                            apiUrl,
-                            model: candidateModel,
-                            variant: variant.name,
-                            messageCount: variant.messages.length,
-                            roles: variant.messages.map(message => message.role),
-                            firstUserPreview: summarizeText(String(variant.messages.find(message => message.role === 'user')?.content || ''), 300)
-                        });
                         const response = await fetch(apiUrl, {
                             method: 'POST',
                             headers: {