openfox 2.0.124 → 2.0.125

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (56) hide show
  1. package/CHANGELOG.md +17 -0
  2. package/README.md +1 -0
  3. package/dist/CHANGELOG.md +17 -0
  4. package/dist/{backend-67OLCS5O.js → backend-4YPEC3J2.js} +2 -2
  5. package/dist/{chat-handler-N43TBYLV.js → chat-handler-HJ2RBGNV.js} +13 -13
  6. package/dist/{chunk-KGFERMCV.js → chunk-7DO4JMGT.js} +6 -6
  7. package/dist/{chunk-UGEFD67O.js → chunk-CTHRGR4O.js} +2 -2
  8. package/dist/{chunk-SYTEO5LA.js → chunk-FNXTJLCZ.js} +197 -8
  9. package/dist/chunk-GZUD5E6E.js +7 -0
  10. package/dist/{chunk-W4SM4FFZ.js → chunk-HOSFRPPI.js} +14 -10
  11. package/dist/{chunk-MIPQOHWS.js → chunk-KAZWU73N.js} +26 -2
  12. package/dist/{chunk-FAL4WO4V.js → chunk-KETVPTMI.js} +3 -3
  13. package/dist/{chunk-62GUO2EG.js → chunk-LJX7KXRU.js} +3 -3
  14. package/dist/{chunk-QJNBQCAF.js → chunk-NBOSTFTS.js} +4 -4
  15. package/dist/{chunk-5CCM655F.js → chunk-O5A3XKE5.js} +18 -9
  16. package/dist/{chunk-AZHWLLPP.js → chunk-PBFGJFGC.js} +28 -10
  17. package/dist/{chunk-VUBFPRKT.js → chunk-T2RDFRNK.js} +56 -49
  18. package/dist/{chunk-A6EZBX62.js → chunk-T3JVYO3Z.js} +16 -6
  19. package/dist/{chunk-JEZN37KV.js → chunk-UAAFZEIS.js} +2 -2
  20. package/dist/{chunk-57UCBFBD.js → chunk-WMAW37GV.js} +9 -9
  21. package/dist/{chunk-RXLAFEZB.js → chunk-X2J6BAU7.js} +4 -4
  22. package/dist/{chunk-3NMLSANH.js → chunk-Y3BSBKHY.js} +31 -17
  23. package/dist/chunk-ZFNBATE3.js +196 -0
  24. package/dist/cli/dev.js +1 -1
  25. package/dist/cli/index.js +1 -1
  26. package/dist/{client-XHJHOFMU.js → client-YHYIWZY7.js} +5 -5
  27. package/dist/{client-pure-QSY7TJGK.js → client-pure-CJKM3ACP.js} +2 -2
  28. package/dist/{compactor-VTATJTPN.js → compactor-7KPBTFZF.js} +5 -5
  29. package/dist/{dynamic-context-KTQCAPTK.js → dynamic-context-5MQWJTJC.js} +2 -2
  30. package/dist/{http-client-N4ASJEKJ.js → http-client-ROVNSHF6.js} +2 -2
  31. package/dist/{launch-Z5D5YFRJ.js → launch-ZHAFPRCP.js} +13 -13
  32. package/dist/{orchestrator-BQBBO4US.js → orchestrator-KKUXIE3Q.js} +12 -12
  33. package/dist/package.json +1 -1
  34. package/dist/plugins-registry.json +6 -0
  35. package/dist/{processor-WTRVYXAO.js → processor-HETUXJFH.js} +12 -12
  36. package/dist/provider/index.d.ts +2 -2
  37. package/dist/{provider-LICG53GG.js → provider-6LUA73HF.js} +6 -6
  38. package/dist/{provider-manager-3QC2MSSR.js → provider-manager-OQTEXZ7X.js} +6 -6
  39. package/dist/{serve-75G5Z7UQ.js → serve-UVHJHYF3.js} +18 -18
  40. package/dist/server/index.d.ts +6 -1
  41. package/dist/server/index.js +17 -17
  42. package/dist/{server-N3VATHIA.js → server-E4DOH4OF.js} +16 -16
  43. package/dist/{service-OURGS2SI.js → service-BRKMLM4C.js} +7 -7
  44. package/dist/{tasks-4SOLTX6A.js → tasks-DEUZ6BNS.js} +7 -7
  45. package/dist/{tools-G44LBKD4.js → tools-AUHL5MPG.js} +10 -10
  46. package/dist/{types-Cr4EgGwK.d.ts → types-BWuVnaBw.d.ts} +3 -1
  47. package/dist/{update-R2TEHD7V.js → update-45ZRDWUB.js} +2 -2
  48. package/dist/web/assets/index-DQ3Wg2OF.js +324 -0
  49. package/dist/web/index.html +1 -1
  50. package/dist/web/sw.js +1 -1
  51. package/dist/{workspace-J3JEGF6U.js → workspace-6CUE26YA.js} +2 -2
  52. package/package.json +1 -1
  53. package/plugins-registry.json +6 -0
  54. package/dist/chunk-SBADOWVC.js +0 -7
  55. package/dist/chunk-VPRB5FNR.js +0 -167
  56. package/dist/web/assets/index-GZTGHRRr.js +0 -324
package/CHANGELOG.md CHANGED
@@ -1,5 +1,22 @@
1
1
  # Changelog
2
2
 
3
+ ## 2.0.125 - 2026-08-21
4
+
5
+ ### Features
6
+
7
+ - **xAI Grok (SuperGrok) plugin** — new curated plugin logs in with your X (xAI) SuperGrok subscription via OAuth, no API key needed
8
+
9
+ ### Enhancements
10
+
11
+ - **Small-context models are flagged** — the model picker and edit-model modal warn when a model's context window sits under 16k tokens
12
+
13
+ ### Bug Fixes
14
+
15
+ - **llama.cpp now honors the reasoning effort** — the chosen level reaches the model (it was silently ignored); 'none' turns thinking off
16
+ - **Symlinked project folders show up in discovery** — directory browser, workspace and plugin listings now include symlinked directories
17
+ - **Ollama requests are fixed end to end** — the native /api/chat endpoint uses the model's full context window so prompts are no longer truncated, unsupported params are dropped, and multi-turn tool calls work again
18
+ - **Thinking mode follows the resolved reasoning effort** — session and sub-agent thinking is no longer force-enabled
19
+
3
20
  ## 2.0.124 - 2026-08-19
4
21
 
5
22
  ### Enhancements
package/README.md CHANGED
@@ -99,6 +99,7 @@ Restart OpenFox after installing or updating a plugin.
99
99
  - To authenticate with a ChatGPT Plus or Pro account, you can install the [`openfox-chatgpt`](https://github.com/arthurlacoste/openfox-chatgpt) plugin.
100
100
  - To authenticate with a Github copilot account, you can install the [`openfox-github-copilot`](https://github.com/JamesDAdams/openfox-github-copilot) plugin.
101
101
  - To authenticate with a Google Antigravity account, you can install the [`openfox-google-antigravity`](https://github.com/JamesDAdams/openfox-google-antigravity) plugin.
102
+ - To authenticate with an X (xAI) SuperGrok subscription, you can install the [`openfox-xai-supergrok`](https://github.com/Olgean-Group/openfox-xai-supergrok) plugin.
102
103
 
103
104
  ## Screenshots
104
105
 
package/dist/CHANGELOG.md CHANGED
@@ -1,5 +1,22 @@
1
1
  # Changelog
2
2
 
3
+ ## 2.0.125 - 2026-08-21
4
+
5
+ ### Features
6
+
7
+ - **xAI Grok (SuperGrok) plugin** — new curated plugin logs in with your X (xAI) SuperGrok subscription via OAuth, no API key needed
8
+
9
+ ### Enhancements
10
+
11
+ - **Small-context models are flagged** — the model picker and edit-model modal warn when a model's context window sits under 16k tokens
12
+
13
+ ### Bug Fixes
14
+
15
+ - **llama.cpp now honors the reasoning effort** — the chosen level reaches the model (it was silently ignored); 'none' turns thinking off
16
+ - **Symlinked project folders show up in discovery** — directory browser, workspace and plugin listings now include symlinked directories
17
+ - **Ollama requests are fixed end to end** — the native /api/chat endpoint uses the model's full context window so prompts are no longer truncated, unsupported params are dropped, and multi-turn tool calls work again
18
+ - **Thinking mode follows the resolved reasoning effort** — session and sub-agent thinking is no longer force-enabled
19
+
3
20
  ## 2.0.124 - 2026-08-19
4
21
 
5
22
  ### Enhancements
@@ -1,10 +1,10 @@
1
1
  import {
2
2
  getBackendCapabilities,
3
3
  getBackendDisplayName
4
- } from "./chunk-AZHWLLPP.js";
4
+ } from "./chunk-PBFGJFGC.js";
5
5
  import "./chunk-YLJ4XMA6.js";
6
6
  export {
7
7
  getBackendCapabilities,
8
8
  getBackendDisplayName
9
9
  };
10
- //# sourceMappingURL=backend-67OLCS5O.js.map
10
+ //# sourceMappingURL=backend-4YPEC3J2.js.map
@@ -2,23 +2,23 @@ import {
2
2
  buildRunChatTurnParams,
3
3
  finalizeTurnCompletion,
4
4
  generateSessionNameForSession
5
- } from "./chunk-W4SM4FFZ.js";
5
+ } from "./chunk-HOSFRPPI.js";
6
6
  import {
7
7
  runChatTurn
8
- } from "./chunk-QJNBQCAF.js";
9
- import "./chunk-5CCM655F.js";
10
- import "./chunk-FAL4WO4V.js";
8
+ } from "./chunk-NBOSTFTS.js";
9
+ import "./chunk-O5A3XKE5.js";
10
+ import "./chunk-KETVPTMI.js";
11
11
  import "./chunk-ZIL6VKLU.js";
12
- import "./chunk-62GUO2EG.js";
12
+ import "./chunk-LJX7KXRU.js";
13
13
  import "./chunk-GIJUWPRQ.js";
14
14
  import "./chunk-EHFIUXWZ.js";
15
15
  import "./chunk-RV5G7QQU.js";
16
16
  import "./chunk-I72ACQ6R.js";
17
17
  import "./chunk-JE7LW7Y6.js";
18
18
  import "./chunk-OAN4BXEW.js";
19
- import "./chunk-3NMLSANH.js";
19
+ import "./chunk-Y3BSBKHY.js";
20
20
  import "./chunk-J7GAZGZT.js";
21
- import "./chunk-UGEFD67O.js";
21
+ import "./chunk-CTHRGR4O.js";
22
22
  import "./chunk-7YP4O4XE.js";
23
23
  import "./chunk-KD4W3I6C.js";
24
24
  import {
@@ -34,18 +34,18 @@ import {
34
34
  import "./chunk-66MG44TF.js";
35
35
  import "./chunk-JURM3RPZ.js";
36
36
  import "./chunk-LCLH6ZUL.js";
37
- import "./chunk-A6EZBX62.js";
37
+ import "./chunk-T3JVYO3Z.js";
38
38
  import "./chunk-K2QKX4ZJ.js";
39
39
  import "./chunk-J2GP3J3X.js";
40
40
  import "./chunk-KBESUA3W.js";
41
41
  import "./chunk-OSXPQSLW.js";
42
42
  import "./chunk-3B232XG4.js";
43
- import "./chunk-SYTEO5LA.js";
43
+ import "./chunk-FNXTJLCZ.js";
44
44
  import "./chunk-GJL3QSMP.js";
45
- import "./chunk-AZHWLLPP.js";
45
+ import "./chunk-PBFGJFGC.js";
46
46
  import "./chunk-HNCM3D7Y.js";
47
- import "./chunk-MIPQOHWS.js";
48
- import "./chunk-VPRB5FNR.js";
47
+ import "./chunk-KAZWU73N.js";
48
+ import "./chunk-ZFNBATE3.js";
49
49
  import "./chunk-N2Q3KYWM.js";
50
50
  import "./chunk-OYBUJIC3.js";
51
51
  import "./chunk-ISE55FOR.js";
@@ -174,4 +174,4 @@ export {
174
174
  startChatSession,
175
175
  stopSessionExecution
176
176
  };
177
- //# sourceMappingURL=chat-handler-N43TBYLV.js.map
177
+ //# sourceMappingURL=chat-handler-HJ2RBGNV.js.map
@@ -1,14 +1,14 @@
1
1
  import {
2
2
  runAgentTurn
3
- } from "./chunk-QJNBQCAF.js";
3
+ } from "./chunk-NBOSTFTS.js";
4
4
  import {
5
5
  executeSubAgent,
6
6
  getToolRegistryForAgent
7
- } from "./chunk-5CCM655F.js";
7
+ } from "./chunk-O5A3XKE5.js";
8
8
  import {
9
9
  TurnMetrics,
10
10
  createMessageStartEvent
11
- } from "./chunk-62GUO2EG.js";
11
+ } from "./chunk-LJX7KXRU.js";
12
12
  import {
13
13
  checkAborted,
14
14
  decodeUtf8,
@@ -27,7 +27,7 @@ import {
27
27
  loadProjectWorkflows,
28
28
  loadUserWorkflows,
29
29
  normalizeWorkflowScope
30
- } from "./chunk-UGEFD67O.js";
30
+ } from "./chunk-CTHRGR4O.js";
31
31
  import {
32
32
  getCurrentContextWindowId,
33
33
  getEventStore
@@ -44,7 +44,7 @@ import {
44
44
  } from "./chunk-KBESUA3W.js";
45
45
  import {
46
46
  LLMError
47
- } from "./chunk-VPRB5FNR.js";
47
+ } from "./chunk-ZFNBATE3.js";
48
48
  import {
49
49
  logger
50
50
  } from "./chunk-K44MW7JJ.js";
@@ -1002,4 +1002,4 @@ export {
1002
1002
  abortRunnerRun,
1003
1003
  launchWorkflowRun
1004
1004
  };
1005
- //# sourceMappingURL=chunk-KGFERMCV.js.map
1005
+ //# sourceMappingURL=chunk-7DO4JMGT.js.map
@@ -1,6 +1,6 @@
1
1
  import {
2
2
  parseDefaultModelSelection
3
- } from "./chunk-A6EZBX62.js";
3
+ } from "./chunk-T3JVYO3Z.js";
4
4
  import {
5
5
  deleteItemFromDir,
6
6
  getDefaultIds,
@@ -1099,4 +1099,4 @@ export {
1099
1099
  isTaskGateError,
1100
1100
  createTasksService
1101
1101
  };
1102
- //# sourceMappingURL=chunk-UGEFD67O.js.map
1102
+ //# sourceMappingURL=chunk-CTHRGR4O.js.map
@@ -3,9 +3,10 @@ import {
3
3
  } from "./chunk-GJL3QSMP.js";
4
4
  import {
5
5
  getBackendCapabilities
6
- } from "./chunk-AZHWLLPP.js";
6
+ } from "./chunk-PBFGJFGC.js";
7
7
  import {
8
- ensureVersionPrefix
8
+ ensureVersionPrefix,
9
+ stripVersionPrefix
9
10
  } from "./chunk-HNCM3D7Y.js";
10
11
  import {
11
12
  buildNonStreamingCreateParams,
@@ -13,15 +14,199 @@ import {
13
14
  getThinking,
14
15
  mapFinishReason,
15
16
  parseToolArguments
16
- } from "./chunk-MIPQOHWS.js";
17
+ } from "./chunk-KAZWU73N.js";
17
18
  import {
19
+ ChatHttpClient,
18
20
  LLMError,
19
21
  OpenAIHttpClient
20
- } from "./chunk-VPRB5FNR.js";
22
+ } from "./chunk-ZFNBATE3.js";
21
23
  import {
22
24
  logger
23
25
  } from "./chunk-K44MW7JJ.js";
24
26
 
27
+ // src/server/llm/ollama-native.ts
28
+ function mapDoneReason(reason) {
29
+ switch (reason) {
30
+ case "length":
31
+ return "length";
32
+ case "tool_calls":
33
+ return "tool_calls";
34
+ case "content_filter":
35
+ return "content_filter";
36
+ default:
37
+ return "stop";
38
+ }
39
+ }
40
+ function stringifyToolArguments(rawArguments) {
41
+ if (typeof rawArguments === "string") return rawArguments;
42
+ if (rawArguments === void 0 || rawArguments === null) return "{}";
43
+ return JSON.stringify(rawArguments);
44
+ }
45
+ var OLLAMA_THINK_LEVELS = /* @__PURE__ */ new Set(["low", "medium", "high", "max"]);
46
+ function toOllamaThink(effort) {
47
+ if (effort === "none") return false;
48
+ if (effort === "xhigh") return "max";
49
+ if (OLLAMA_THINK_LEVELS.has(effort)) return effort;
50
+ return true;
51
+ }
52
+ function toNativeMessage(message) {
53
+ const raw = message;
54
+ const toolCalls = raw["tool_calls"];
55
+ if (!Array.isArray(toolCalls) || toolCalls.length === 0) {
56
+ return raw;
57
+ }
58
+ return {
59
+ ...raw,
60
+ tool_calls: toolCalls.map((toolCall) => {
61
+ const fn = toolCall.function;
62
+ if (!fn || typeof fn.arguments !== "string") return toolCall;
63
+ try {
64
+ return {
65
+ ...toolCall,
66
+ function: { ...fn, arguments: JSON.parse(fn.arguments) }
67
+ };
68
+ } catch {
69
+ return toolCall;
70
+ }
71
+ })
72
+ };
73
+ }
74
+ function buildOllamaChatRequest(params) {
75
+ const options = {};
76
+ if (params.temperature !== void 0) options["temperature"] = params.temperature;
77
+ if (params.top_p !== void 0) options["top_p"] = params.top_p;
78
+ if (params.top_k !== void 0) options["top_k"] = params.top_k;
79
+ if (params.max_tokens !== void 0) options["num_predict"] = params.max_tokens;
80
+ if (params["num_ctx"] !== void 0) options["num_ctx"] = params["num_ctx"];
81
+ if (params["stop"]) options["stop"] = params["stop"];
82
+ if (params["frequency_penalty"] !== void 0) options["frequency_penalty"] = params["frequency_penalty"];
83
+ if (params["presence_penalty"] !== void 0) options["presence_penalty"] = params["presence_penalty"];
84
+ if (params["seed"] !== void 0) options["seed"] = params["seed"];
85
+ const body = {
86
+ model: params.model,
87
+ messages: params.messages.map(toNativeMessage),
88
+ stream: Boolean(params.stream),
89
+ ...params.tools && params.tools.length > 0 ? { tools: params.tools } : {},
90
+ ...Object.keys(options).length > 0 ? { options } : {}
91
+ };
92
+ if (params["response_format"] && typeof params["response_format"] === "object") {
93
+ const type = params["response_format"].type;
94
+ if (type === "json_object" || type === "json") {
95
+ body["format"] = "json";
96
+ }
97
+ }
98
+ const kwargs = params.chat_template_kwargs;
99
+ if (kwargs && typeof kwargs["enable_thinking"] === "boolean") {
100
+ body["think"] = kwargs["enable_thinking"];
101
+ }
102
+ if (params.reasoning_effort) {
103
+ body["think"] = toOllamaThink(params.reasoning_effort);
104
+ }
105
+ return body;
106
+ }
107
+ function nativeToolCallToOpenAIResponse(toolCall, index) {
108
+ const fn = toolCall.function ?? {};
109
+ return {
110
+ id: toolCall.id ?? `call_${index}`,
111
+ type: "function",
112
+ function: {
113
+ name: fn.name ?? "",
114
+ arguments: stringifyToolArguments(fn.arguments)
115
+ }
116
+ };
117
+ }
118
+ function nativeToolCallToOpenAIStream(toolCall, index) {
119
+ const fn = toolCall.function ?? {};
120
+ return {
121
+ index: typeof fn.index === "number" ? fn.index : index,
122
+ id: toolCall.id ?? `call_${index}`,
123
+ function: {
124
+ name: fn.name ?? "",
125
+ arguments: stringifyToolArguments(fn.arguments)
126
+ }
127
+ };
128
+ }
129
+ function parseOllamaChatResponse(data) {
130
+ const msg = data.message ?? {};
131
+ const toolCalls = (msg.tool_calls ?? []).map(nativeToolCallToOpenAIResponse);
132
+ const promptTokens = typeof data.prompt_eval_count === "number" ? data.prompt_eval_count : 0;
133
+ const completionTokens = typeof data.eval_count === "number" ? data.eval_count : 0;
134
+ const message = {
135
+ content: msg.content ?? ""
136
+ };
137
+ if (msg.reasoning) message["reasoning_content"] = msg.reasoning;
138
+ if (msg.thinking) message["thinking"] = msg.thinking;
139
+ if (toolCalls.length > 0) message["tool_calls"] = toolCalls;
140
+ return {
141
+ id: `chatcmpl-${Date.now()}`,
142
+ choices: [
143
+ {
144
+ finish_reason: mapDoneReason(data.done_reason),
145
+ message
146
+ }
147
+ ],
148
+ usage: {
149
+ prompt_tokens: promptTokens,
150
+ completion_tokens: completionTokens,
151
+ total_tokens: promptTokens + completionTokens
152
+ }
153
+ };
154
+ }
155
+ function parseOllamaChatChunk(data) {
156
+ const msg = data.message ?? {};
157
+ const toolCalls = (msg.tool_calls ?? []).map(nativeToolCallToOpenAIStream);
158
+ const delta = {};
159
+ if (msg.content) delta["content"] = msg.content;
160
+ if (msg.reasoning) delta["reasoning_content"] = msg.reasoning;
161
+ if (msg.thinking) delta["thinking"] = msg.thinking;
162
+ if (toolCalls.length > 0) delta["tool_calls"] = toolCalls;
163
+ const promptTokens = typeof data.prompt_eval_count === "number" ? data.prompt_eval_count : void 0;
164
+ const completionTokens = typeof data.eval_count === "number" ? data.eval_count : void 0;
165
+ return {
166
+ id: `chatcmpl-${Date.now()}`,
167
+ choices: [
168
+ {
169
+ delta,
170
+ finish_reason: data.done ? mapDoneReason(data.done_reason) : null
171
+ }
172
+ ],
173
+ ...promptTokens !== void 0 ? {
174
+ usage: {
175
+ prompt_tokens: promptTokens,
176
+ completion_tokens: completionTokens ?? 0,
177
+ total_tokens: promptTokens + (completionTokens ?? 0)
178
+ }
179
+ } : {}
180
+ };
181
+ }
182
+ var OllamaHttpClient = class extends ChatHttpClient {
183
+ baseURL;
184
+ constructor(options) {
185
+ super();
186
+ this.baseURL = options.baseURL;
187
+ }
188
+ buildRequest(params) {
189
+ return {
190
+ url: `${this.baseURL}/api/chat`,
191
+ headers: {
192
+ "Content-Type": "application/json"
193
+ },
194
+ body: JSON.stringify(buildOllamaChatRequest(params))
195
+ };
196
+ }
197
+ parseNonStreaming(data) {
198
+ return parseOllamaChatResponse(data);
199
+ }
200
+ parseStreamLine(trimmed) {
201
+ try {
202
+ return parseOllamaChatChunk(JSON.parse(trimmed));
203
+ } catch (error) {
204
+ logger.warn("Failed to parse Ollama stream chunk", { data: trimmed, error });
205
+ return null;
206
+ }
207
+ }
208
+ };
209
+
25
210
  // src/server/llm/client.ts
26
211
  function extractContentFromBlocks(blocks) {
27
212
  let text = "";
@@ -39,12 +224,16 @@ function extractContentFromBlocks(blocks) {
39
224
  }
40
225
  return { text, thinking };
41
226
  }
42
- function createLLMClient(config, initialBackend = "unknown") {
227
+ function createLLMClient(config, initialBackend = config.llm.backend ?? "unknown") {
43
228
  const baseURL = ensureVersionPrefix(config.llm.baseUrl);
44
229
  const httpClient = new OpenAIHttpClient({
45
230
  baseURL,
46
231
  apiKey: config.llm.apiKey ?? "not-needed"
47
232
  });
233
+ const ollamaHttpClient = new OllamaHttpClient({
234
+ baseURL: stripVersionPrefix(baseURL)
235
+ });
236
+ const httpFor = (b) => b === "ollama" ? ollamaHttpClient : httpClient;
48
237
  let model = config.llm.model;
49
238
  let profile = getModelProfile(model);
50
239
  let backend = initialBackend;
@@ -103,7 +292,7 @@ function createLLMClient(config, initialBackend = "unknown") {
103
292
  capabilities,
104
293
  ...buildExtraParams(resolvedEffort)
105
294
  });
106
- const httpResponse = await httpClient.createChatCompletion(
295
+ const httpResponse = await httpFor(backend).createChatCompletion(
107
296
  createParams,
108
297
  {
109
298
  signal: request.signal
@@ -180,7 +369,7 @@ function createLLMClient(config, initialBackend = "unknown") {
180
369
  }
181
370
  }, 100);
182
371
  const streamSignal = request.signal ? AbortSignal.any([request.signal, idleTimeoutController.signal]) : idleTimeoutController.signal;
183
- const stream = httpClient.createChatCompletionStream(streamingParams, {
372
+ const stream = httpFor(backend).createChatCompletionStream(streamingParams, {
184
373
  signal: streamSignal
185
374
  });
186
375
  let fullContent = "";
@@ -317,4 +506,4 @@ function createLLMClient(config, initialBackend = "unknown") {
317
506
  export {
318
507
  createLLMClient
319
508
  };
320
- //# sourceMappingURL=chunk-SYTEO5LA.js.map
509
+ //# sourceMappingURL=chunk-FNXTJLCZ.js.map
@@ -0,0 +1,7 @@
1
+ // src/constants.ts
2
+ var VERSION = "2.0.125";
3
+
4
+ export {
5
+ VERSION
6
+ };
7
+ //# sourceMappingURL=chunk-GZUD5E6E.js.map
@@ -131,7 +131,8 @@ function resolveSessionProvider(session, providerManager, effective) {
131
131
  providerId: provider.id,
132
132
  baseUrl: provider.url,
133
133
  ...provider.apiKey ? { apiKey: provider.apiKey } : {},
134
- model: effectiveModel
134
+ model: effectiveModel,
135
+ ...provider.backend ? { backend: provider.backend } : {}
135
136
  };
136
137
  }
137
138
  async function generateSessionNameForSession(sessionId, userMessage, deps, signal) {
@@ -161,14 +162,17 @@ async function generateSessionNameForSession(sessionId, userMessage, deps, signa
161
162
  client = deps.getLLMClient();
162
163
  client.setModel(providerConfig.model);
163
164
  } else if (providerConfig) {
164
- const { createLLMClient } = await import("./client-XHJHOFMU.js");
165
- client = createLLMClient({
166
- llm: {
167
- baseUrl: providerConfig.baseUrl,
168
- model: providerConfig.model,
169
- ...providerConfig.apiKey ? { apiKey: providerConfig.apiKey } : {}
170
- }
171
- });
165
+ const { createLLMClient } = await import("./client-YHYIWZY7.js");
166
+ client = createLLMClient(
167
+ {
168
+ llm: {
169
+ baseUrl: providerConfig.baseUrl,
170
+ model: providerConfig.model,
171
+ ...providerConfig.apiKey ? { apiKey: providerConfig.apiKey } : {}
172
+ }
173
+ },
174
+ providerConfig.backend ?? "unknown"
175
+ );
172
176
  } else if (deps.getLLMClient) {
173
177
  client = deps.getLLMClient();
174
178
  } else {
@@ -212,4 +216,4 @@ export {
212
216
  needsNameGeneration,
213
217
  generateSessionNameForSession
214
218
  };
215
- //# sourceMappingURL=chunk-W4SM4FFZ.js.map
219
+ //# sourceMappingURL=chunk-HOSFRPPI.js.map
@@ -617,11 +617,35 @@ async function buildChatCompletionCreateParams(model, request, profile, capabili
617
617
  ;
618
618
  params["top_k"] = topK;
619
619
  }
620
+ if (capabilities.supportsNumCtx && request.modelSettings?.numCtx) {
621
+ ;
622
+ params["num_ctx"] = request.modelSettings.numCtx;
623
+ }
620
624
  const resolvedEffort = reasoningEffort ?? request.reasoningEffort;
621
625
  const queryParams = request.modelSettings?.queryParams;
622
626
  const hasQueryParams = queryParams && Object.keys(queryParams).length > 0;
623
627
  const hasExplicitModelSettings = hasQueryParams || !!request.modelSettings?.chatTemplateKwargs;
624
- if (hasQueryParams) {
628
+ if (capabilities.routesEffortViaChatTemplateKwargs) {
629
+ let userKwargs = request.modelSettings?.chatTemplateKwargs;
630
+ if (hasQueryParams) {
631
+ const { chat_template_kwargs: userKwargsFromQP, ...restQueryParams } = queryParams;
632
+ Object.assign(params, restQueryParams);
633
+ if (userKwargsFromQP) {
634
+ userKwargs = userKwargsFromQP;
635
+ }
636
+ }
637
+ const kwargs = { ...userKwargs ?? {} };
638
+ if (resolvedEffort === "none") {
639
+ delete kwargs["reasoning_effort"];
640
+ kwargs["enable_thinking"] = false;
641
+ } else if (resolvedEffort && kwargs["reasoning_effort"] === void 0) {
642
+ kwargs["reasoning_effort"] = resolvedEffort;
643
+ }
644
+ if (Object.keys(kwargs).length > 0) {
645
+ ;
646
+ params["chat_template_kwargs"] = kwargs;
647
+ }
648
+ } else if (hasQueryParams) {
625
649
  Object.assign(params, queryParams);
626
650
  if (resolvedEffort) {
627
651
  ;
@@ -711,4 +735,4 @@ export {
711
735
  buildStreamingCreateParams,
712
736
  mapFinishReason
713
737
  };
714
- //# sourceMappingURL=chunk-MIPQOHWS.js.map
738
+ //# sourceMappingURL=chunk-KAZWU73N.js.map
@@ -105,7 +105,7 @@ function resolveAgentDef(sessionManager, sessionId) {
105
105
  }
106
106
  async function buildCachedPrompt(sessionManager, sessionId, agentDef, modelName) {
107
107
  const { instructionContent, skills } = await loadSessionContext(sessionManager, sessionId);
108
- const { getToolRegistryForAgent } = await import("./tools-G44LBKD4.js");
108
+ const { getToolRegistryForAgent } = await import("./tools-AUHL5MPG.js");
109
109
  const tools = getToolRegistryForAgent(agentDef, sessionId).definitions;
110
110
  const toolFingerprint = getToolFingerprint(tools);
111
111
  const allAgents = await loadAllAgentsDefault(sessionManager.getProjectWorkdir(sessionId));
@@ -124,7 +124,7 @@ async function buildCachedPrompt(sessionManager, sessionId, agentDef, modelName)
124
124
  async function computeSessionHash(sessionManager, sessionId, modelName) {
125
125
  const { instructionContent, skills } = await loadSessionContext(sessionManager, sessionId);
126
126
  const agentDef = await resolveAgentDef(sessionManager, sessionId);
127
- const { getToolRegistryForAgent } = await import("./tools-G44LBKD4.js");
127
+ const { getToolRegistryForAgent } = await import("./tools-AUHL5MPG.js");
128
128
  const tools = getToolRegistryForAgent(agentDef, sessionId).definitions;
129
129
  const toolFingerprint = getToolFingerprint(tools);
130
130
  return computeDynamicContextHash(instructionContent, skills, toolFingerprint, modelName);
@@ -148,4 +148,4 @@ export {
148
148
  computeSessionHash,
149
149
  applyDynamicContext
150
150
  };
151
- //# sourceMappingURL=chunk-FAL4WO4V.js.map
151
+ //# sourceMappingURL=chunk-KETVPTMI.js.map
@@ -6,10 +6,10 @@ import {
6
6
  } from "./chunk-GJL3QSMP.js";
7
7
  import {
8
8
  getBackendCapabilities
9
- } from "./chunk-AZHWLLPP.js";
9
+ } from "./chunk-PBFGJFGC.js";
10
10
  import {
11
11
  buildModelParams
12
- } from "./chunk-MIPQOHWS.js";
12
+ } from "./chunk-KAZWU73N.js";
13
13
  import {
14
14
  logger
15
15
  } from "./chunk-K44MW7JJ.js";
@@ -518,4 +518,4 @@ export {
518
518
  createChatDoneEvent,
519
519
  consumeStreamGenerator
520
520
  };
521
- //# sourceMappingURL=chunk-62GUO2EG.js.map
521
+ //# sourceMappingURL=chunk-LJX7KXRU.js.map
@@ -4,12 +4,12 @@ import {
4
4
  getToolRegistryForAgent,
5
5
  processEventsForConversation,
6
6
  runTopLevelAgentLoop
7
- } from "./chunk-5CCM655F.js";
7
+ } from "./chunk-O5A3XKE5.js";
8
8
  import {
9
9
  buildCachedPrompt,
10
10
  computeDynamicContextHash,
11
11
  getToolFingerprint
12
- } from "./chunk-FAL4WO4V.js";
12
+ } from "./chunk-KETVPTMI.js";
13
13
  import {
14
14
  getAllInstructions,
15
15
  getEnabledSkillMetadata
@@ -18,7 +18,7 @@ import {
18
18
  TurnMetrics,
19
19
  createChatDoneEvent,
20
20
  createMessageStartEvent
21
- } from "./chunk-62GUO2EG.js";
21
+ } from "./chunk-LJX7KXRU.js";
22
22
  import {
23
23
  buildAgentReminder,
24
24
  buildAgentSmallReminder
@@ -329,4 +329,4 @@ export {
329
329
  runChatTurn,
330
330
  runAgentTurn
331
331
  };
332
- //# sourceMappingURL=chunk-QJNBQCAF.js.map
332
+ //# sourceMappingURL=chunk-NBOSTFTS.js.map