fix(newos4): send num_ctx to Ollama, and make one auto-connect failure read as one line

Two things I had flagged as "worth your attention" and one leftover nit.

- num_ctx was never sent. Ollama's server-side default context is 2k-4k depending on the build, so
  a system prompt + 20 messages + up to 24k chars of tool results could be truncated server-side,
  which is exactly the "it forgot what it just read" class of report. Every Ollama request now
  carries options.num_ctx (default 16384, overridable via webx11_ollama_numctx, floor 512); the
  OpenAI-compatible branch is untouched because num_ctx is not one of its parameters.
- Automatic connect prints one outcome instead of one line per endpoint tried (7 -> 3 on a dead
  endpoint). Interactive `ollamaconnect` keeps the per-endpoint detail.
- write_file's failure message no longer echoes a trailing slash.

Verification:
- test-ollama-numctx.js (7 asserts): generate and chat payloads carry num_ctx=16384, a stored
  override is used, junk/too-small values fall back, and the /v1/chat/completions branch does not
  receive it. RED on HEAD: TypeError reading 'num_ctx' of undefined. Suite 20/20.
- Live v4.6.117, capturing the real request: {"num_ctx":16384} on /api/generate with
  model gemma4:31b-cloud, reply "OK".
- Live v4.6.118, dead endpoint (127.0.0.1:9, fallback disabled): automatic -> 3 lines
  ("connecting on first use…", "Connecting to dead…", "Auto-connect failed — …"); manual
  connect() -> 4 lines including "endpoint http://127.0.0.1:9 failed", so the suppression is
  conditional, not blanket.

1 files changed, 28 insertions, 10 deletions

diff --git a/modules/core/core-bundle.js b/modules/core/core-bundle.js
index 1a92ef9..fcf1505 100644
--- a/modules/core/core-bundle.js
+++ b/modules/core/core-bundle.js
@@ -1290,6 +1290,19 @@ class OllamaManager {
 
     // ─── Provider management ─────────────────────────────────────────────────
 
+    // Ollama's server-side default context is small (2k-4k depending on the build). The agent
+    // sends a system prompt, up to 20 messages and up to 24k chars of tool results, so without an
+    // explicit num_ctx the server silently truncates the prompt — the model "forgets" what it was
+    // just told. Send it on every Ollama request; the OpenAI-compatible branch has its own params.
+    _ctxTokens() {
+        const n = Number(this.numCtx || (typeof localStorage !== 'undefined' ? localStorage.getItem('webx11_ollama_numctx') : 0));
+        return Number.isFinite(n) && n >= 512 ? Math.floor(n) : 16384;
+    }
+
+    _ollamaOptions(extra) {
+        return Object.assign({ num_ctx: this._ctxTokens() }, extra || {});
+    }
+
     _loadProviders() {
         const stored = this._getSecret('webx11_llm_providers');
         if (stored) {
@@ -1506,7 +1519,7 @@ class OllamaManager {
                         break;
                     } catch (error) {
                         this.terminal.log('WARN', 'Ollama', provider.name + ' endpoint ' + ep + ' failed: ' + error.message);
-                        this.terminal.print(`${provider.name} endpoint ${ep} failed: ${error.message}`, 'warning');
+                        if (!this._quietConnect) this.terminal.print(`${provider.name} endpoint ${ep} failed: ${error.message}`, 'warning');
                     }
                 }
             } else {
@@ -1544,7 +1557,7 @@ class OllamaManager {
                     success = true;
                 } catch (error) {
                     this.terminal.log('WARN', 'Ollama', provider.name + ' failed: ' + error.message);
-                    this.terminal.print(`${provider.name} failed: ${error.message}`, 'warning');
+                    if (!this._quietConnect) this.terminal.print(`${provider.name} failed: ${error.message}`, 'warning');
                 }
             }
 
@@ -1554,7 +1567,7 @@ class OllamaManager {
         // If provider-based connect failed, try bare endpoint fallback (legacy)
         for (const ep of allEndpoints) {
             try {
-                this.terminal.print(`Connecting (legacy fallback) to ${ep}...`, 'info');
+                if (!this._quietConnect) this.terminal.print(`Connecting (legacy fallback) to ${ep}...`, 'info');
                 const response = await fetch(`${ep}/api/tags`, {
                     method: 'GET', headers: { 'Content-Type': 'application/json' }
                 });
@@ -1580,7 +1593,7 @@ class OllamaManager {
                 return true;
             } catch (error) {
                 this.terminal.log('WARN', 'Ollama', 'Endpoint ' + ep + ' failed: ' + error.message);
-                this.terminal.print(`Endpoint ${ep} failed: ${error.message}`, 'warning');
+                if (!this._quietConnect) this.terminal.print(`Endpoint ${ep} failed: ${error.message}`, 'warning');
             }
         }
 
@@ -1622,6 +1635,10 @@ class OllamaManager {
         this.terminal.print('Ollama: connecting on first use…', 'info');
         let timedOut = false;
         try {
+            // One failure should read as one failure: connect() prints a line per endpoint it
+            // tries, which made an auto-connect failure seven lines of noise. Keep the detail in
+            // the log, print only the outcome for an automatic attempt.
+            this._quietConnect = true;
             this._connecting = this.connect();
             // connect() uses a bare fetch(), so a blackholed endpoint would hang this promise
             // forever and — because the promise is shared — every later message with it.
@@ -1635,6 +1652,7 @@ class OllamaManager {
             // A throwing terminal.log() must not leave _connecting poisoned: log defensively.
             try { this.terminal.log('WARN', 'Ollama', 'Auto-connect failed: ' + (e && e.message ? e.message : e)); } catch (_) {}
         } finally {
+            this._quietConnect = false;
             this._connecting = null;
         }
         if (this.connected && savedModel && !(this.models || []).some(m => m.name === savedModel)) {
@@ -1706,7 +1724,7 @@ class OllamaManager {
                 const response = await fetch(`${provider.endpoint}/api/generate`, {
                     method: 'POST',
                     headers: { 'Content-Type': 'application/json' },
-                    body: JSON.stringify({ model: useModel, prompt: prompt, stream: this.streamEnabled }),
+                    body: JSON.stringify({ model: useModel, prompt: prompt, stream: this.streamEnabled, options: this._ollamaOptions() }),
                     signal
                 });
                 if (!response.ok) throw new Error(`HTTP ${response.status}`);
@@ -1876,7 +1894,7 @@ class OllamaManager {
                 const response = await fetch(`${provider.endpoint}/api/generate`, {
                     method: 'POST',
                     headers: { 'Content-Type': 'application/json' },
-                    body: JSON.stringify({ model: useModel, prompt: prompt, stream: this.streamEnabled }),
+                    body: JSON.stringify({ model: useModel, prompt: prompt, stream: this.streamEnabled, options: this._ollamaOptions() }),
                     signal
                 });
                 if (!response.ok) throw new Error(`HTTP ${response.status}`);
@@ -2053,7 +2071,7 @@ class OllamaManager {
 
         try {
             if (provider.type === 'ollama') {
-                const body = { model: useModel, messages, stream: false };
+                const body = { model: useModel, messages, stream: false, options: this._ollamaOptions() };
                 if (tools && tools.length > 0) body.tools = tools;
                 const t0 = Date.now();
                 const response = await fetch(`${provider.endpoint}/api/chat`, {
@@ -2286,7 +2304,7 @@ ${toolDesc}
 
 Only output XML function calls for tool tasks. If no tool is needed, answer normally.` };
                 const msgsWithTools = [toolContext, ...messages];
-                const body = { model: useModel, messages: msgsWithTools, stream: false, options: { temperature: 0.1 } };
+                const body = { model: useModel, messages: msgsWithTools, stream: false, options: this._ollamaOptions({ temperature: 0.1 }) };
 
                 const response = await fetch(`${this.endpoint}/api/chat`, {
                     method: 'POST', headers: { 'Content-Type': 'application/json' },
@@ -2337,7 +2355,7 @@ Only output XML function calls for tool tasks. If no tool is needed, answer norm
         // Original Ollama fallback
         this.terminal.print(`Model doesn't support tool calling. Retrying /api/chat without tools...`, 'warning');
         try {
-            const body = { model: useModel, messages, stream: false };
+            const body = { model: useModel, messages, stream: false, options: this._ollamaOptions() };
             const response = await fetch(`${this.endpoint}/api/chat`, {
                 method: 'POST', headers: { 'Content-Type': 'application/json' },
                 body: JSON.stringify(body),
@@ -11084,7 +11102,7 @@ buildSystemPrompt() {
                                 const why = (typeof this.terminal.fs._isVirtualPath === 'function' && this.terminal.fs._isVirtualPath(wp))
                                     ? 'read-only virtual path'
                                     : 'parent directory missing or target is a directory';
-                                return `Write FAILED: ${wp} — ${why}; nothing was written`;
+                                return `Write FAILED: ${String(wp).replace(/\/+$/, '') || '/'} — ${why}; nothing was written`;
                             }
                             return `File written: ${wp} (${content.length} chars)`;
                         }