v1.11.3: fix ctx_max capture via /props endpoint

- llama-server does not emit n_ctx in timings (confirmed empirically);
  dead code at inference.ts:479 and compaction.ts:300 never fired
- New model-context.ts: cached fetch of /upstream/<model>/props
  with positive-cache (no TTL) and 60s negative-cache
- Wired into all 4 ctx_max write sites: 3 in inference.ts
  (executeToolPhase, finalizeCompletion, runCapHitSummary) and
  1 in compaction.ts (summary row INSERT)
- AbortController 3s timeout, lenient parsing with sensible defaults
- 12 new vitest cases for the cache module (59 total)
- 7 historical assistant rows backfilled manually (see notes)

apps/server/src/services/inference.ts

@@ -22,6 +22,7 @@ import { PathScopeError, resolveProjectRoot } from './path_guard.js';
 import { maybeAutoNameChat } from './auto_name.js';
 import { getAgentById } from './agents.js';
 import * as compaction from './compaction.js';
+import * as modelContext from './model-context.js';
 import type { Broker } from './broker.js';
 
 const BASE_SYSTEM_PROMPT = (projectPath: string) =>
@@ -138,9 +139,6 @@ interface ChatCompletionChunk {
     completion_tokens?: number;
     total_tokens?: number;
   };
-  timings?: {
-    n_ctx?: number;
-  };
 }
 
 export interface InferenceContext {
@@ -339,7 +337,6 @@ interface StreamResult {
   toolCalls: ToolCall[];
   promptTokens: number | null;
   completionTokens: number | null;
-  nCtx: number | null;
 }
 
 interface StreamOptions {
@@ -454,7 +451,6 @@ async function streamCompletion(
   let finishReason: string | null = null;
   let promptTokens: number | null = null;
   let completionTokens: number | null = null;
-  let nCtx: number | null = null;
   const toolCallsBuffer = new Map<number, { id: string; name: string; argsText: string }>();
 
   for await (const line of sseLines(res.body)) {
@@ -476,9 +472,11 @@ async function streamCompletion(
         completionTokens = parsed.usage.completion_tokens;
       }
     }
-    if (parsed.timings && typeof parsed.timings.n_ctx === 'number') {
-      nCtx = parsed.timings.n_ctx;
-    }
+    // v1.11.3: removed dead `parsed.timings.n_ctx` read. llama-server's
+    // streaming completion does NOT emit n_ctx in timings (verified
+    // empirically); the authoritative source is llama-swap's
+    // /upstream/<model>/props endpoint, fetched per-turn via
+    // model-context.getModelContext() at the finalization sites below.
 
     const choice = parsed.choices?.[0];
     if (!choice) continue;
@@ -564,7 +562,7 @@ async function streamCompletion(
     toolCalls.push({ id: t.id || `call_${toolCalls.length}`, name: t.name, args });
   }
 
-  return { finishReason, content, toolCalls, promptTokens, completionTokens, nCtx };
+  return { finishReason, content, toolCalls, promptTokens, completionTokens };
 }
 
 async function executeToolCall(
@@ -781,7 +779,14 @@ async function executeToolPhase(
   projectRoot: string
 ): Promise<void> {
   const { sessionId, chatId, assistantMessageId, toolsUsed, signal } = args;
-  const { content, toolCalls, promptTokens, completionTokens, nCtx } = result;
+  const { content, toolCalls, promptTokens, completionTokens } = result;
+
+  // v1.11.3: ctx_max comes from llama-swap /upstream/<model>/props, not the
+  // streaming completion (which doesn't emit n_ctx). getModelContext caches
+  // the positive lookup for the process lifetime, so this is a single Map
+  // hit after the first invocation per model.
+  const mctx = await modelContext.getModelContext(session.model);
+  const nCtx = mctx?.n_ctx ?? null;
 
   const [updated] = await ctx.sql<
     { tokens_used: number | null; ctx_used: number | null; ctx_max: number | null; finished_at: string | null }[]
@@ -917,7 +922,11 @@ async function finalizeCompletion(
   session: Session
 ): Promise<void> {
   const { sessionId, chatId, assistantMessageId } = args;
-  const { content, finishReason, promptTokens, completionTokens, nCtx } = result;
+  const { content, finishReason, promptTokens, completionTokens } = result;
+
+  // v1.11.3: see executeToolPhase for the rationale.
+  const mctx = await modelContext.getModelContext(session.model);
+  const nCtx = mctx?.n_ctx ?? null;
 
   const [updated] = await ctx.sql<
     { tokens_used: number | null; ctx_used: number | null; ctx_max: number | null; finished_at: string | null }[]
@@ -1150,6 +1159,9 @@ async function runCapHitSummary(
   // even on a partial / failed summary the chat history shows where the
   // budget was hit.
   if (summaryOk && result) {
+    // v1.11.3: see executeToolPhase for the rationale.
+    const mctx = await modelContext.getModelContext(session.model);
+    const nCtx = mctx?.n_ctx ?? null;
     const [updated] = await ctx.sql<
       { tokens_used: number | null; ctx_used: number | null; ctx_max: number | null; finished_at: string | null }[]
     >`
@@ -1158,7 +1170,7 @@ async function runCapHitSummary(
           status = 'complete',
           tokens_used = ${result.completionTokens},
           ctx_used = ${result.promptTokens},
-          ctx_max = ${result.nCtx},
+          ctx_max = ${nCtx},
           finished_at = clock_timestamp()
       WHERE id = ${assistantMessageId}
       RETURNING tokens_used, ctx_used, ctx_max, finished_at