feat(newos4): configurable tool budgets, per-model defaults, real tool telemetry

Closes STEER-20261102-201234-7200 (read_file capped at 6000 while the schema
promised 32000).

Budgets (module-aiassist.js + new modules/core/model-profiles.js):
- one resolver, precedence: call arg -> loop profile -> model profile ->
  ai-config -> code default. Keys: read-page-chars, read-result-chars,
  tool-context-budget, fetch-chars.
- model classes tiny/small/mid/large keyed by id then by size token; `mid`
  reproduces the old numbers so nothing moves for a mid model.
- _clipToolResult now clips the `content` field, never the serialised JSON
  envelope: clipping the envelope deleted truncated/next_offset, so a 16k page
  lost its continuation pointer (the reported symptom).
- read_file's schema no longer promises a number the model class may not give.

Telemetry:
- _recordToolCall keeps sizes, duration, ok and a 120-char arg preview instead
  of the whole payload; ai-stats prints calls / ok% / avg ms / KB / truncation
  from localStorage webx11_tool_stats.

Server (api/proxy-api.py):
- /api/extract honours max_chars (default 12000, cap 100000) and returns
  content_chars + truncated; the bs4 branch is no longer uncapped while the
  regex fallback was silently cut at 10 000.

BEHAVIOUR CHANGE: num_ctx for a model with no explicit setting now follows the
class default (16384 -> 65536 for cloud/large, 8192 for tiny). Cloud models bill
on tokens used, so cost is unchanged; local models get a bigger KV cache.

Two build traps, both now documented in AGENTS.md:
- model-profiles.js must be the FIRST entry in CORE_FILES and must export the
  MODEL_PROFILES global (a TDZ typeof throws; without the global every budget
  silently falls back to the code default).
- a budget key needs the setter branch AND the constructor loader at
  module-aiassist.js:49 - persisting without restoring looks fine until reload.

Verified: test-tool-context-bound 23/23 (RED before: E1-E4 + TypeError on
_toolLimit); live bundle read_page 32000 / result 12000 / budget 48000 /
fetch 12000 and _ctxTokens 65536 for gemma4:31b-cloud; /api/extract 28970 ->
500 chars with truncated true; ai-config write+restore probe 6/6.

1 files changed, 45 insertions, 13 deletions

diff --git a/modules/core/ollama-manager.js b/modules/core/ollama-manager.js
index 7dcecc3..1201280 100644
--- a/modules/core/ollama-manager.js
+++ b/modules/core/ollama-manager.js
@@ -2,6 +2,31 @@
 // Refactored to support multiple LLM providers with auto-fallback
 // Class name kept as OllamaManager for backward compatibility
 class OllamaManager {
+
+    async _httpError(response) {
+        // Surface what the server actually said. "HTTP 429" tells the user nothing; the body
+        // carries e.g. "Demo limit reached. Login or sign up for more." (2026-10-28)
+        let detail = '';
+        try {
+            const text = await response.text();
+            if (text) {
+                try {
+                    const j = JSON.parse(text);
+                    detail = (j && (j.error || j.message)) || '';
+                    if (j && j.needed_usd != null && j.balance_usd != null) {
+                        detail += ` (balance $${j.balance_usd}, need $${j.needed_usd})`;
+                    }
+                } catch (_) {
+                    detail = text.slice(0, 200);
+                }
+            }
+        } catch (_) { /* body unreadable or already consumed */ }
+        if (/^\s*</.test(detail)) detail = '';   // an HTML error page is not a message
+        const err = new Error(detail ? `${detail} (HTTP ${response.status})` : `HTTP ${response.status}`);
+        err.status = response.status;
+        return err;
+    }
+
     constructor(terminal) {
         this.terminal = terminal;
         // --- Backward-compat properties (reflected from active provider) ---
@@ -61,7 +86,14 @@ class OllamaManager {
     // 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;
+        if (Number.isFinite(n) && n >= 512) return Math.floor(n);
+        // No explicit setting: take the model class default (large/cloud 65536, mid 32768, …)
+        try {
+            const mp = this.modelProfiles
+                || (typeof MODEL_PROFILES !== 'undefined' ? MODEL_PROFILES : null);
+            if (mp && mp.profileFor && this.currentModel) return mp.profileFor(this.currentModel).ctx;
+        } catch (e) {}
+        return 16384;
     }
 
     _ollamaOptions(extra) {
@@ -261,7 +293,7 @@ class OllamaManager {
                         const response = await fetch(`${ep}/api/tags`, {
                             method: 'GET', headers: { 'Content-Type': 'application/json' }
                         });
-                        if (!response.ok) throw new Error(`HTTP ${response.status}`);
+                        if (!response.ok) throw await this._httpError(response);
                         const data = await response.json();
                         this.models = data.models || [];
                         this.connected = true;
@@ -298,7 +330,7 @@ class OllamaManager {
                     const response = await fetch(`${provider.endpoint}/v1/models`, {
                         method: 'GET', headers
                     });
-                    if (!response.ok) throw new Error(`HTTP ${response.status}`);
+                    if (!response.ok) throw await this._httpError(response);
                     const data = await response.json();
                     this.models = (data.data || data.models || []).map(m => ({
                         name: m.id || m.name,
@@ -336,7 +368,7 @@ class OllamaManager {
                 const response = await fetch(`${ep}/api/tags`, {
                     method: 'GET', headers: { 'Content-Type': 'application/json' }
                 });
-                if (!response.ok) throw new Error(`HTTP ${response.status}`);
+                if (!response.ok) throw await this._httpError(response);
                 const data = await response.json();
                 this.models = data.models || [];
                 this.connected = true;
@@ -492,7 +524,7 @@ class OllamaManager {
                     body: JSON.stringify({ model: useModel, prompt: prompt, stream: this.streamEnabled, options: this._ollamaOptions() }),
                     signal
                 });
-                if (!response.ok) throw new Error(`HTTP ${response.status}`);
+                if (!response.ok) throw await this._httpError(response);
 
                 if (this.streamEnabled) {
                     const reader = response.body.getReader();
@@ -549,7 +581,7 @@ class OllamaManager {
                     }),
                     signal
                 });
-                if (!response.ok) throw new Error(`HTTP ${response.status}`);
+                if (!response.ok) throw await this._httpError(response);
 
                 if (this.streamEnabled) {
                     const reader = response.body.getReader();
@@ -662,7 +694,7 @@ class OllamaManager {
                     body: JSON.stringify({ model: useModel, prompt: prompt, stream: this.streamEnabled, options: this._ollamaOptions() }),
                     signal
                 });
-                if (!response.ok) throw new Error(`HTTP ${response.status}`);
+                if (!response.ok) throw await this._httpError(response);
 
                 let fullResponse = '';
                 if (this.streamEnabled) {
@@ -725,7 +757,7 @@ class OllamaManager {
                     }),
                     signal
                 });
-                if (!response.ok) throw new Error(`HTTP ${response.status}`);
+                if (!response.ok) throw await this._httpError(response);
 
                 let fullResponse = '';
                 if (this.streamEnabled) {
@@ -850,7 +882,7 @@ class OllamaManager {
                     if (response.status === 400) {
                         return await this.chatWithToolsFallback(messages, tools, useModel, signal);
                     }
-                    throw new Error(`HTTP ${response.status}`);
+                    throw await this._httpError(response);
                 }
                 const data = await response.json();
                 this.terminal.log('INFO', 'Ollama', `chatWithTools round-trip: ${networkTime}s (model: ${useModel}, provider: ${provider.name})`);
@@ -900,7 +932,7 @@ class OllamaManager {
                 });
                 const networkTime = Math.round((Date.now() - t0) / 1000);
                 if (!response.ok) {
-                    throw new Error(`HTTP ${response.status}`);
+                    throw await this._httpError(response);
                 }
                 const data = await response.json();
                 this.terminal.log('INFO', 'Ollama', `chatWithTools round-trip: ${networkTime}s (model: ${useModel}, provider: ${provider.name})`);
@@ -1039,7 +1071,7 @@ class OllamaManager {
                     body: JSON.stringify(body),
                     signal
                 });
-                if (!response.ok) throw new Error(`HTTP ${response.status}`);
+                if (!response.ok) throw await this._httpError(response);
                 const data = await response.json();
                 const text = data.choices?.[0]?.message?.content || '';
                 const latency = Date.now() - t0;
@@ -1075,7 +1107,7 @@ Only output XML function calls for tool tasks. If no tool is needed, answer norm
                     method: 'POST', headers: { 'Content-Type': 'application/json' },
                     body: JSON.stringify(body), signal
                 });
-                if (!response.ok) throw new Error(`HTTP ${response.status}`);
+                if (!response.ok) throw await this._httpError(response);
                 const data = await response.json();
                 const text = (data.message?.content || '').trim();
                 const latency = Date.now() - t0;
@@ -1126,7 +1158,7 @@ Only output XML function calls for tool tasks. If no tool is needed, answer norm
                 body: JSON.stringify(body),
                 signal
             });
-            if (!response.ok) throw new Error(`HTTP ${response.status}`);
+            if (!response.ok) throw await this._httpError(response);
             const data = await response.json();
             const text = data.message?.content || data.response || '';
             const latency = Date.now() - t0;