@goodandready/dsh-ollama-cloud-provider 0.1.17 → 0.1.19

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.
package/CHANGELOG.md CHANGED
@@ -1,5 +1,29 @@
1
1
  # Changelog
2
2
 
3
+ ## 0.1.19 — 2026-11-05
4
+
5
+ - Added configurable native generation defaults: temperature, top_k, top_p,
6
+ min_p, seed, repeat_penalty, repeat_last_n, num_ctx, keep_alive and stop.
7
+ Request options override config; Ollama-only fields are not sent to the
8
+ OpenAI transport. Added settings fields for all values.
9
+ - Normalized thinking ladders: boolean switches become off/high, level lists
10
+ are ordered and deduplicated, and known families get a default effort
11
+ (GLM-5.x/Kimi K3 max, Kimi K2/DeepSeek V4/MiniMax M3 high, GPT-OSS medium,
12
+ Nemotron low) only when that level is offered.
13
+ - Added generation/reasoning tests; 112/112 tests.
14
+
15
+ ## 0.1.18 — 2026-11-05
16
+
17
+ - Added local/self-hosted Ollama mode: an explicit empty `apiKeyEnv` means no
18
+ authentication. Native and OpenAI requests omit `Authorization`, the key pool
19
+ returns a synthetic empty credential, and usage/balance requests are skipped.
20
+ - `/api/show` probing enriches local model discovery with context window,
21
+ vision, tools and thinking capabilities; successful probes are cached for five
22
+ minutes and cleared with the catalog cache or explicitly.
23
+ - Catalog discovery now falls back to `/api/tags` ids when `/v1/models` is
24
+ empty and adopts `/api/show` metadata for local endpoints.
25
+ - Added local-mode, probe-cache and no-auth tests; 109/109 tests.
26
+
3
27
  ## 0.1.17 — 2026-11-05
4
28
 
5
29
  - Added Ollama web capabilities behind `ctx.web`: a search provider and a fetch
package/README.md CHANGED
@@ -52,6 +52,22 @@ Select it in the profile:
52
52
  fetchProvider: ollama-cloud
53
53
  ```
54
54
 
55
+ ## Local Ollama
56
+
57
+ Set `apiKeyEnv` to an empty string and point `apiBase` at the local or
58
+ self-hosted Ollama server. No `Authorization` header is sent, and cloud
59
+ usage/balance reads are skipped.
60
+
61
+ ```yaml
62
+ - id: llm-ollama-cloud
63
+ config:
64
+ apiKeyEnv: ''
65
+ apiBase: http://127.0.0.1:11434
66
+ ```
67
+
68
+ Local model discovery uses `/api/tags` and enriches each model through
69
+ `/api/show`; successful capability probes are cached for five minutes.
70
+
55
71
  ## Architecture
56
72
 
57
73
  ```mermaid
@@ -121,17 +137,17 @@ npm pack
121
137
  Install the published package:
122
138
 
123
139
  ```bash
124
- dsh plugin --profile web add @goodandready/dsh-ollama-cloud-provider@0.1.17
140
+ dsh plugin --profile web add @goodandready/dsh-ollama-cloud-provider@0.1.19
125
141
  # DSH Desktop:
126
- dsh plugin --profile desktop add @goodandready/dsh-ollama-cloud-provider@0.1.17
142
+ dsh plugin --profile desktop add @goodandready/dsh-ollama-cloud-provider@0.1.19
127
143
  ```
128
144
 
129
145
  Or install the packed tarball into a profile:
130
146
 
131
147
  ```bash
132
- dsh plugin --profile web add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.17.tgz
148
+ dsh plugin --profile web add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.19.tgz
133
149
  # DSH Desktop:
134
- dsh plugin --profile desktop add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.17.tgz
150
+ dsh plugin --profile desktop add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.19.tgz
135
151
  ```
136
152
 
137
153
  Restart dsh (or DSH Desktop), then pick **Ollama Cloud** in the model picker.
package/README.ru.md CHANGED
@@ -103,17 +103,17 @@ npm pack
103
103
  Установка опубликованного пакета из npm:
104
104
 
105
105
  ```bash
106
- dsh plugin --profile web add @goodandready/dsh-ollama-cloud-provider@0.1.17
106
+ dsh plugin --profile web add @goodandready/dsh-ollama-cloud-provider@0.1.19
107
107
  # DSH Desktop:
108
- dsh plugin --profile desktop add @goodandready/dsh-ollama-cloud-provider@0.1.17
108
+ dsh plugin --profile desktop add @goodandready/dsh-ollama-cloud-provider@0.1.19
109
109
  ```
110
110
 
111
111
  Либо установка из упакованного tarball:
112
112
 
113
113
  ```bash
114
- dsh plugin --profile web add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.17.tgz
114
+ dsh plugin --profile web add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.19.tgz
115
115
  # DSH Desktop:
116
- dsh plugin --profile desktop add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.17.tgz
116
+ dsh plugin --profile desktop add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.19.tgz
117
117
  ```
118
118
 
119
119
  После установки перезапустите dsh (или DSH Desktop) и выберите **Ollama Cloud** в списке моделей.
package/README.zh.md CHANGED
@@ -102,17 +102,17 @@ npm pack
102
102
  从 npm 安装已发布版本:
103
103
 
104
104
  ```bash
105
- dsh plugin --profile web add @goodandready/dsh-ollama-cloud-provider@0.1.17
105
+ dsh plugin --profile web add @goodandready/dsh-ollama-cloud-provider@0.1.19
106
106
  # DSH 桌面端:
107
- dsh plugin --profile desktop add @goodandready/dsh-ollama-cloud-provider@0.1.17
107
+ dsh plugin --profile desktop add @goodandready/dsh-ollama-cloud-provider@0.1.19
108
108
  ```
109
109
 
110
110
  或通过打包文件安装:
111
111
 
112
112
  ```bash
113
- dsh plugin --profile web add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.17.tgz
113
+ dsh plugin --profile web add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.19.tgz
114
114
  # DSH 桌面端:
115
- dsh plugin --profile desktop add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.17.tgz
115
+ dsh plugin --profile desktop add /absolute/path/to/goodandready-dsh-ollama-cloud-provider-0.1.19.tgz
116
116
  ```
117
117
 
118
118
  安装完成后重启 dsh 或 DSH 桌面端,在模型列表中选择 **Ollama Cloud** 即可。
package/lib/client.js CHANGED
@@ -122,6 +122,20 @@ window.__ModuleLoader__.load({
122
122
  streamIdleTimeoutMs: "Stream idle timeout (ms)",
123
123
  maxOutputTokens: "Output limit (0 = provider default)",
124
124
  contextWindow: "Context window override",
125
+ generation: "Generation defaults",
126
+ generationHint: "Native Ollama defaults. Empty fields are not sent; request options still win.",
127
+ temperature: "Temperature",
128
+ topK: "Top K",
129
+ topP: "Top P",
130
+ minP: "Min P",
131
+ seed: "Seed",
132
+ repeatPenalty: "Repeat penalty",
133
+ repeatLastN: "Repeat last N",
134
+ numCtx: "num_ctx",
135
+ keepAlive: "Keep alive",
136
+ keepAliveHint: "Ollama keep_alive, for example 5m or -1.",
137
+ stop: "Stop sequences",
138
+ stopHint: "Comma-separated; sent as the native stop list.",
125
139
  sidebar: "Sidebar",
126
140
  serviceStatus: "Service status",
127
141
  serviceStatusHint: "Checks the public Ollama Cloud tag endpoint without using your key or quota.",
@@ -295,6 +309,20 @@ window.__ModuleLoader__.load({
295
309
  streamIdleTimeoutMs: "流空闲超时(毫秒)",
296
310
  maxOutputTokens: "输出上限(0 = 提供方默认)",
297
311
  contextWindow: "上下文窗口覆盖",
312
+ generation: "生成默认值",
313
+ generationHint: "Ollama 原生默认值;空字段不发送,请求参数优先。",
314
+ temperature: "Temperature",
315
+ topK: "Top K",
316
+ topP: "Top P",
317
+ minP: "Min P",
318
+ seed: "Seed",
319
+ repeatPenalty: "Repeat penalty",
320
+ repeatLastN: "Repeat last N",
321
+ numCtx: "num_ctx",
322
+ keepAlive: "Keep alive",
323
+ keepAliveHint: "Ollama keep_alive,例如 5m 或 -1。",
324
+ stop: "Stop sequences",
325
+ stopHint: "逗号分隔,作为原生 stop 列表发送。",
298
326
  sidebar: "侧边栏",
299
327
  serviceStatus: "服务状态",
300
328
  serviceStatusHint: "检查 Ollama Cloud 公共 tag 端点,不使用密钥或额度。",
@@ -1486,6 +1514,9 @@ window.__ModuleLoader__.load({
1486
1514
  });
1487
1515
  }
1488
1516
  // src/client/settings-draft.ts
1517
+ function textOf(value) {
1518
+ return value === null || value === void 0 ? "" : String(value);
1519
+ }
1489
1520
  function draftOf(state) {
1490
1521
  const allIds = state.models.map((model) => model.id);
1491
1522
  return {
@@ -1497,8 +1528,18 @@ window.__ModuleLoader__.load({
1497
1528
  transport: state.config.transport,
1498
1529
  requestTimeoutMs: String(state.config.requestTimeoutMs),
1499
1530
  streamIdleTimeoutMs: String(state.config.streamIdleTimeoutMs),
1500
- maxOutputTokens: state.config.maxOutputTokens === null ? "" : String(state.config.maxOutputTokens),
1501
- contextWindow: state.config.contextWindow === null ? "" : String(state.config.contextWindow),
1531
+ maxOutputTokens: textOf(state.config.maxOutputTokens),
1532
+ contextWindow: textOf(state.config.contextWindow),
1533
+ temperature: textOf(state.config.temperature),
1534
+ topK: textOf(state.config.topK),
1535
+ topP: textOf(state.config.topP),
1536
+ minP: textOf(state.config.minP),
1537
+ seed: textOf(state.config.seed),
1538
+ repeatPenalty: textOf(state.config.repeatPenalty),
1539
+ repeatLastN: textOf(state.config.repeatLastN),
1540
+ numCtx: textOf(state.config.numCtx),
1541
+ keepAlive: state.config.keepAlive ?? "",
1542
+ stop: (state.config.stop ?? []).join(", "),
1502
1543
  showSidebarQuota: state.config.showSidebarQuota,
1503
1544
  planLabel: state.config.planLabel,
1504
1545
  activeAccount: state.config.activeAccount,
@@ -1518,6 +1559,16 @@ window.__ModuleLoader__.load({
1518
1559
  if (!Number.isInteger(value) || value < 0) throw new Error(`Invalid number: ${text}`);
1519
1560
  return value;
1520
1561
  }
1562
+ function boundedNumber(text, min, max, integer) {
1563
+ const trimmed = text.trim();
1564
+ if (trimmed === "") return null;
1565
+ const value = Number(trimmed);
1566
+ if (!Number.isFinite(value) || value < min || value > max || integer && !Number.isInteger(value)) throw new Error(`Invalid number: ${text}`);
1567
+ return value;
1568
+ }
1569
+ function stopList(text) {
1570
+ return text.split(/[\n,]/).map((entry) => entry.trim()).filter((entry) => entry !== "");
1571
+ }
1521
1572
  function patchOf(draft) {
1522
1573
  return {
1523
1574
  apiKeyEnv: draft.apiKeyEnv.trim(),
@@ -1530,6 +1581,16 @@ window.__ModuleLoader__.load({
1530
1581
  streamIdleTimeoutMs: numberOrNull(draft.streamIdleTimeoutMs) ?? 3e5,
1531
1582
  maxOutputTokens: numberOrNull(draft.maxOutputTokens),
1532
1583
  contextWindow: numberOrNull(draft.contextWindow),
1584
+ temperature: boundedNumber(draft.temperature, 0, 2, false),
1585
+ topK: boundedNumber(draft.topK, 1, 1e6, true),
1586
+ topP: boundedNumber(draft.topP, 0, 1, false),
1587
+ minP: boundedNumber(draft.minP, 0, 1, false),
1588
+ seed: boundedNumber(draft.seed, 0, 4294967295, true),
1589
+ repeatPenalty: boundedNumber(draft.repeatPenalty, 0, 2, false),
1590
+ repeatLastN: boundedNumber(draft.repeatLastN, -1, 1e6, true),
1591
+ numCtx: boundedNumber(draft.numCtx, 1, 1e7, true),
1592
+ keepAlive: draft.keepAlive.trim(),
1593
+ stop: stopList(draft.stop),
1533
1594
  showSidebarQuota: draft.showSidebarQuota,
1534
1595
  planLabel: draft.planLabel.trim(),
1535
1596
  activeAccount: draft.activeAccount.trim(),
@@ -2508,6 +2569,84 @@ window.__ModuleLoader__.load({
2508
2569
  })
2509
2570
  ]
2510
2571
  }),
2572
+ /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("section", {
2573
+ className: "ocl-card",
2574
+ children: [
2575
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)("h3", {
2576
+ className: "ocl-card-title",
2577
+ children: t("generation")
2578
+ }),
2579
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)("p", {
2580
+ className: "ocl-hint",
2581
+ children: t("generationHint")
2582
+ }),
2583
+ /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
2584
+ className: "ocl-grid",
2585
+ children: [
2586
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)(Field, {
2587
+ label: t("temperature"),
2588
+ type: "number",
2589
+ value: draft.temperature,
2590
+ onChange: (value) => update({ temperature: value })
2591
+ }),
2592
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)(Field, {
2593
+ label: t("topK"),
2594
+ type: "number",
2595
+ value: draft.topK,
2596
+ onChange: (value) => update({ topK: value })
2597
+ }),
2598
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)(Field, {
2599
+ label: t("topP"),
2600
+ type: "number",
2601
+ value: draft.topP,
2602
+ onChange: (value) => update({ topP: value })
2603
+ }),
2604
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)(Field, {
2605
+ label: t("minP"),
2606
+ type: "number",
2607
+ value: draft.minP,
2608
+ onChange: (value) => update({ minP: value })
2609
+ }),
2610
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)(Field, {
2611
+ label: t("seed"),
2612
+ type: "number",
2613
+ value: draft.seed,
2614
+ onChange: (value) => update({ seed: value })
2615
+ }),
2616
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)(Field, {
2617
+ label: t("repeatPenalty"),
2618
+ type: "number",
2619
+ value: draft.repeatPenalty,
2620
+ onChange: (value) => update({ repeatPenalty: value })
2621
+ }),
2622
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)(Field, {
2623
+ label: t("repeatLastN"),
2624
+ type: "number",
2625
+ value: draft.repeatLastN,
2626
+ onChange: (value) => update({ repeatLastN: value })
2627
+ }),
2628
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)(Field, {
2629
+ label: t("numCtx"),
2630
+ type: "number",
2631
+ value: draft.numCtx,
2632
+ onChange: (value) => update({ numCtx: value })
2633
+ }),
2634
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)(Field, {
2635
+ label: t("keepAlive"),
2636
+ hint: t("keepAliveHint"),
2637
+ value: draft.keepAlive,
2638
+ onChange: (value) => update({ keepAlive: value })
2639
+ }),
2640
+ /* @__PURE__ */ (0, react_jsx_runtime.jsx)(Field, {
2641
+ label: t("stop"),
2642
+ hint: t("stopHint"),
2643
+ value: draft.stop,
2644
+ onChange: (value) => update({ stop: value })
2645
+ })
2646
+ ]
2647
+ })
2648
+ ]
2649
+ }),
2511
2650
  /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("section", {
2512
2651
  className: "ocl-card",
2513
2652
  children: [/* @__PURE__ */ (0, react_jsx_runtime.jsx)("h3", {
package/lib/index.d.ts CHANGED
@@ -58,6 +58,17 @@ interface Config {
58
58
  contextWindowOverrides?: Record<string, number>;
59
59
  /** Per-turn output cap; `0` omits `num_predict`/`max_tokens`. */
60
60
  maxOutputTokens?: number;
61
+ /** Optional Ollama native generation defaults. */
62
+ temperature?: number;
63
+ topK?: number;
64
+ topP?: number;
65
+ minP?: number;
66
+ seed?: number;
67
+ repeatPenalty?: number;
68
+ repeatLastN?: number;
69
+ numCtx?: number;
70
+ keepAlive?: string;
71
+ stop?: string[];
61
72
  /** Extra accounts for rotation. */
62
73
  accounts?: AccountConfig[];
63
74
  /** Preferred account per model glob; others remain fallbacks. */
@@ -94,6 +105,16 @@ interface ResolvedConfig {
94
105
  contextWindow?: number;
95
106
  contextWindowOverrides: Readonly<Record<string, number>>;
96
107
  maxOutputTokens?: number;
108
+ temperature?: number;
109
+ topK?: number;
110
+ topP?: number;
111
+ minP?: number;
112
+ seed?: number;
113
+ repeatPenalty?: number;
114
+ repeatLastN?: number;
115
+ numCtx?: number;
116
+ keepAlive?: string;
117
+ stop?: readonly string[];
97
118
  accounts: readonly AccountConfig[];
98
119
  modelAccountRules: readonly ModelAccountRule[];
99
120
  stickySession: boolean;
@@ -439,6 +460,23 @@ interface EffectiveModel extends OllamaCloudModelMeta {
439
460
  }
440
461
  declare function effectiveModel(id: string, config: ResolvedConfig): EffectiveModel;
441
462
  declare function describeModel(meta: EffectiveModel, config: ResolvedConfig): string;
463
+ interface ModelProbe {
464
+ readonly contextWindow?: number;
465
+ readonly input: readonly ('text' | 'image')[];
466
+ readonly reasoning: readonly string[];
467
+ readonly reasoningToggle: boolean;
468
+ readonly tools: boolean;
469
+ readonly available: boolean;
470
+ }
471
+ /** Probe one local model via `/api/show`, caching successful answers for five minutes. */
472
+ declare function probeModel(config: ResolvedConfig, model: string, fetchImpl?: typeof fetch): Promise<ModelProbe | undefined>;
473
+ /** Clear all probe entries, or only entries for one endpoint. */
474
+ declare function clearModelProbeCache(baseURL?: string): void;
475
+ declare function modelProbeCacheInfo(): Array<{
476
+ baseURL: string;
477
+ model: string;
478
+ at: number;
479
+ }>;
442
480
  declare const name = "llm-ollama-cloud";
443
481
  declare const inject: string[];
444
482
  /**
@@ -447,5 +485,5 @@ declare const inject: string[];
447
485
  * without a restart.
448
486
  */
449
487
  declare function apply(ctx: Context, config: Config): void;
450
- export { type CommandDeps, Config, type Config as OllamaCloudConfig, OllamaCloudAdapter, type OllamaCloudAdapterDeps, PROVIDER, type ResolvedConfig, type ResolvedKey, type RouteDeps, type TransportId, type UsageReport, type UsageResult, type UsageWindow, apply, applyCommands, applyRoutes, classifyResponseError, clearCatalogCache, createKeyPool, createKeyResolver, createUsageReader, describeModel, effectiveModel, inject, name, normalizeBalance, normalizeUsage, resolveConfig };
488
+ export { type CommandDeps, Config, type Config as OllamaCloudConfig, OllamaCloudAdapter, type OllamaCloudAdapterDeps, PROVIDER, type ResolvedConfig, type ResolvedKey, type RouteDeps, type TransportId, type UsageReport, type UsageResult, type UsageWindow, apply, applyCommands, applyRoutes, classifyResponseError, clearCatalogCache, clearModelProbeCache, createKeyPool, createKeyResolver, createUsageReader, describeModel, effectiveModel, inject, modelProbeCacheInfo, name, normalizeBalance, normalizeUsage, probeModel, resolveConfig };
451
489
  //# sourceMappingURL=index.d.ts.map
package/lib/index.js CHANGED
@@ -50,7 +50,7 @@ const DEFAULT_API_BASE = "https://ollama.com";
50
50
  const DEFAULT_OPENAI_BASE = "https://ollama.com/v1";
51
51
  const DEFAULT_USAGE_URL = "https://ollama.com/api/usage";
52
52
  const DEFAULT_BALANCE_URL = "https://ollama.com/api/balance";
53
- const DEFAULT_USER_AGENT = "dsh-ollama-cloud-provider/0.1.17";
53
+ const DEFAULT_USER_AGENT = "dsh-ollama-cloud-provider/0.1.19";
54
54
  const DEFAULT_CONTEXT_WINDOW = 131072;
55
55
  const DEFAULT_MAX_TOKENS = 32768;
56
56
  const DEFAULT_GENERATE_MAX_TOKENS = 131072;
@@ -76,6 +76,16 @@ const Config = z.object({
76
76
  contextWindow: markVolatile(z.number().min(1).max(1e7).step(1)),
77
77
  contextWindowOverrides: markVolatile(z.dict(z.number().min(1).max(1e7).step(1)).default({})),
78
78
  maxOutputTokens: markVolatile(z.number().min(0).max(1e7).step(1)),
79
+ temperature: markVolatile(z.number().min(0).max(2)),
80
+ topK: markVolatile(z.number().min(1).max(1e6).step(1)),
81
+ topP: markVolatile(z.number().min(0).max(1)),
82
+ minP: markVolatile(z.number().min(0).max(1)),
83
+ seed: markVolatile(z.number().min(0).max(4294967295).step(1)),
84
+ repeatPenalty: markVolatile(z.number().min(0).max(2)),
85
+ repeatLastN: markVolatile(z.number().min(-1).max(1e6).step(1)),
86
+ numCtx: markVolatile(z.number().min(1).max(1e7).step(1)),
87
+ keepAlive: markVolatile(z.string()),
88
+ stop: markVolatile(z.array(z.string())),
79
89
  accounts: markVolatile(z.array(z.object({
80
90
  label: z.string(),
81
91
  apiKeyEnv: z.string().role("credential-ref"),
@@ -109,6 +119,18 @@ function unwrapVolatileConfig(config) {
109
119
  function positiveInt(value, fallback) {
110
120
  return typeof value === "number" && Number.isFinite(value) && value > 0 ? Math.trunc(value) : fallback;
111
121
  }
122
+ function numberInRange(value, min, max) {
123
+ return typeof value === "number" && Number.isFinite(value) && value >= min && value <= max ? value : void 0;
124
+ }
125
+ function integerInRange(value, min, max) {
126
+ if (typeof value !== "number" || !Number.isInteger(value) || value < min || value > max) return void 0;
127
+ return value;
128
+ }
129
+ function nonEmptyStrings(value) {
130
+ if (!Array.isArray(value)) return void 0;
131
+ const out = value.filter((entry) => typeof entry === "string" && entry.trim() !== "");
132
+ return out.length === 0 ? void 0 : out;
133
+ }
112
134
  function recordOf(value) {
113
135
  return typeof value === "object" && value !== null && !Array.isArray(value) ? value : {};
114
136
  }
@@ -158,8 +180,17 @@ function ruleList(value) {
158
180
  function resolveConfig(config) {
159
181
  const apiKey = typeof config.apiKey === "string" && config.apiKey.trim() !== "" ? config.apiKey.trim() : void 0;
160
182
  const visibleModels = Array.isArray(config.visibleModels) ? config.visibleModels.filter((id) => typeof id === "string" && id !== "") : [];
183
+ const temperature = numberInRange(config.temperature, 0, 2);
184
+ const topK = integerInRange(config.topK, 1, 1e6);
185
+ const topP = numberInRange(config.topP, 0, 1);
186
+ const minP = numberInRange(config.minP, 0, 1);
187
+ const seed = integerInRange(config.seed, 0, 4294967295);
188
+ const repeatPenalty = numberInRange(config.repeatPenalty, 0, 2);
189
+ const repeatLastN = integerInRange(config.repeatLastN, -1, 1e6);
190
+ const numCtx = integerInRange(config.numCtx, 1, 1e7);
191
+ const stop = nonEmptyStrings(config.stop);
161
192
  return {
162
- apiKeyEnv: typeof config.apiKeyEnv === "string" && config.apiKeyEnv.trim() !== "" ? config.apiKeyEnv.trim() : DEFAULT_API_KEY_ENV,
193
+ apiKeyEnv: typeof config.apiKeyEnv === "string" ? config.apiKeyEnv.trim() : DEFAULT_API_KEY_ENV,
163
194
  ...apiKey === void 0 ? {} : { apiKey },
164
195
  apiBase: typeof config.apiBase === "string" && config.apiBase !== "" ? config.apiBase : DEFAULT_API_BASE,
165
196
  openAiBase: typeof config.openAiBase === "string" && config.openAiBase !== "" ? config.openAiBase : DEFAULT_OPENAI_BASE,
@@ -176,6 +207,16 @@ function resolveConfig(config) {
176
207
  ...typeof config.contextWindow === "number" && Number.isInteger(config.contextWindow) && config.contextWindow > 0 ? { contextWindow: config.contextWindow } : {},
177
208
  contextWindowOverrides: numberRecord(config.contextWindowOverrides),
178
209
  ...typeof config.maxOutputTokens === "number" && Number.isInteger(config.maxOutputTokens) && config.maxOutputTokens >= 0 ? { maxOutputTokens: config.maxOutputTokens } : {},
210
+ ...temperature === void 0 ? {} : { temperature },
211
+ ...topK === void 0 ? {} : { topK },
212
+ ...topP === void 0 ? {} : { topP },
213
+ ...minP === void 0 ? {} : { minP },
214
+ ...seed === void 0 ? {} : { seed },
215
+ ...repeatPenalty === void 0 ? {} : { repeatPenalty },
216
+ ...repeatLastN === void 0 ? {} : { repeatLastN },
217
+ ...numCtx === void 0 ? {} : { numCtx },
218
+ ...typeof config.keepAlive === "string" && config.keepAlive.trim() !== "" ? { keepAlive: config.keepAlive.trim() } : {},
219
+ ...stop === void 0 ? {} : { stop },
179
220
  accounts: Array.isArray(config.accounts) ? config.accounts : [],
180
221
  modelAccountRules: ruleList(config.modelAccountRules),
181
222
  stickySession: config.stickySession !== false,
@@ -682,6 +723,113 @@ const GENERATED_MODELS = {
682
723
  }
683
724
  }
684
725
  };
726
+ // src/routing.ts
727
+ function trim(url) {
728
+ return url.replace(/\/+$/, "");
729
+ }
730
+ function rootFor(config) {
731
+ return trim(config.apiBase);
732
+ }
733
+ function modelsUrlFor(config) {
734
+ return `${rootFor(config)}/v1/models`;
735
+ }
736
+ function tagsUrlFor(config) {
737
+ return `${rootFor(config)}/api/tags`;
738
+ }
739
+ /** Endpoint for one model and one wire protocol. */
740
+ function endpointFor(transport, config) {
741
+ return transport === "openai-chat" ? `${trim(config.openAiBase)}/chat/completions` : `${rootFor(config)}/api/chat`;
742
+ }
743
+ // src/model-probe.ts
744
+ const PROBE_TTL_MS = 3e5;
745
+ const cache$2 = /* @__PURE__ */ new Map();
746
+ function keyOf(config, model) {
747
+ return `${rootFor(config)}|${model}`;
748
+ }
749
+ function isLocalEndpoint(config) {
750
+ if (config.apiKeyEnv === "") return true;
751
+ try {
752
+ const host = new URL(config.apiBase).hostname;
753
+ return host === "127.0.0.1" || host === "localhost" || host === "::1" || host === "[::1]";
754
+ } catch {
755
+ return false;
756
+ }
757
+ }
758
+ function contextOf(info) {
759
+ let best;
760
+ for (const [key, value] of Object.entries(info)) {
761
+ if (!key.endsWith(".context_length")) continue;
762
+ if (typeof value !== "number" || !Number.isFinite(value) || value <= 0) continue;
763
+ best = best === void 0 ? value : Math.max(best, value);
764
+ }
765
+ return best === void 0 ? void 0 : Math.trunc(best);
766
+ }
767
+ function probeOf(body) {
768
+ if (typeof body !== "object" || body === null || Array.isArray(body)) return void 0;
769
+ const record = body;
770
+ const info = typeof record.model_info === "object" && record.model_info !== null ? record.model_info : {};
771
+ const capabilities = Array.isArray(record.capabilities) ? record.capabilities.filter((entry) => typeof entry === "string") : [];
772
+ const thinking = capabilities.includes("thinking");
773
+ const vision = capabilities.includes("vision");
774
+ const contextWindow = contextOf(info);
775
+ const input = vision ? ["text", "image"] : ["text"];
776
+ return {
777
+ ...contextWindow === void 0 ? {} : { contextWindow },
778
+ input,
779
+ reasoning: thinking ? ["off", "high"] : [],
780
+ reasoningToggle: thinking,
781
+ tools: capabilities.includes("tools"),
782
+ available: true
783
+ };
784
+ }
785
+ /** Probe one local model via `/api/show`, caching successful answers for five minutes. */
786
+ async function probeModel(config, model, fetchImpl = fetch) {
787
+ if (!isLocalEndpoint(config)) return void 0;
788
+ const key = keyOf(config, model);
789
+ const hit = cache$2.get(key);
790
+ if (hit !== void 0 && Date.now() - hit.at < PROBE_TTL_MS) return hit.value;
791
+ try {
792
+ const response = await fetchImpl(`${rootFor(config)}/api/show`, {
793
+ method: "POST",
794
+ headers: {
795
+ accept: "application/json",
796
+ "content-type": "application/json",
797
+ "user-agent": config.userAgent
798
+ },
799
+ body: JSON.stringify({ model }),
800
+ signal: AbortSignal.timeout(1e4)
801
+ });
802
+ if (!response.ok) return void 0;
803
+ const value = probeOf(await response.json());
804
+ if (value === void 0) return void 0;
805
+ cache$2.set(key, {
806
+ at: Date.now(),
807
+ value
808
+ });
809
+ return value;
810
+ } catch {
811
+ return;
812
+ }
813
+ }
814
+ /** Clear all probe entries, or only entries for one endpoint. */
815
+ function clearModelProbeCache(baseURL) {
816
+ if (baseURL === void 0) {
817
+ cache$2.clear();
818
+ return;
819
+ }
820
+ const prefix = `${baseURL.replace(/\/+$/, "")}|`;
821
+ for (const key of cache$2.keys()) if (key.startsWith(prefix)) cache$2.delete(key);
822
+ }
823
+ function modelProbeCacheInfo() {
824
+ return [...cache$2.entries()].map(([key, entry]) => {
825
+ const separator = key.indexOf("|");
826
+ return {
827
+ baseURL: key.slice(0, separator),
828
+ model: key.slice(separator + 1),
829
+ at: entry.at
830
+ };
831
+ });
832
+ }
685
833
  // src/pricing.ts
686
834
  /** `gpt-oss:120b` → `gpt-oss`; used as the second lookup key. */
687
835
  function baseIdOf(id) {
@@ -756,22 +904,73 @@ function formatUsd(value) {
756
904
  if (Math.abs(value) >= .01) return `$${value.toFixed(3)}`;
757
905
  return `$${value.toFixed(4)}`;
758
906
  }
759
- // src/routing.ts
760
- function trim(url) {
761
- return url.replace(/\/+$/, "");
762
- }
763
- function rootFor(config) {
764
- return trim(config.apiBase);
765
- }
766
- function modelsUrlFor(config) {
767
- return `${rootFor(config)}/v1/models`;
768
- }
769
- function tagsUrlFor(config) {
770
- return `${rootFor(config)}/api/tags`;
771
- }
772
- /** Endpoint for one model and one wire protocol. */
773
- function endpointFor(transport, config) {
774
- return transport === "openai-chat" ? `${trim(config.openAiBase)}/chat/completions` : `${rootFor(config)}/api/chat`;
907
+ // src/reasoning.ts
908
+ const EFFORT_ORDER = [
909
+ "off",
910
+ "minimal",
911
+ "low",
912
+ "medium",
913
+ "high",
914
+ "xhigh",
915
+ "max"
916
+ ];
917
+ const FAMILY_DEFAULTS = [
918
+ {
919
+ prefix: "glm-5.3",
920
+ effort: "max"
921
+ },
922
+ {
923
+ prefix: "glm-5.2",
924
+ effort: "max"
925
+ },
926
+ {
927
+ prefix: "kimi-k3",
928
+ effort: "max"
929
+ },
930
+ {
931
+ prefix: "kimi-k2",
932
+ effort: "high"
933
+ },
934
+ {
935
+ prefix: "deepseek-v4",
936
+ effort: "high"
937
+ },
938
+ {
939
+ prefix: "minimax-m3",
940
+ effort: "high"
941
+ },
942
+ {
943
+ prefix: "gpt-oss",
944
+ effort: "medium"
945
+ },
946
+ {
947
+ prefix: "nemotron",
948
+ effort: "low"
949
+ }
950
+ ];
951
+ function normalizeEfforts(levels) {
952
+ const seen = /* @__PURE__ */ new Set();
953
+ for (const level of levels) {
954
+ const clean = level.trim().toLowerCase();
955
+ if (clean !== "") seen.add(clean);
956
+ }
957
+ const ordered = EFFORT_ORDER.filter((level) => seen.has(level));
958
+ const extra = [...seen].filter((level) => !EFFORT_ORDER.includes(level)).sort();
959
+ return [...ordered, ...extra];
960
+ }
961
+ function defaultEffortFor(id, levels) {
962
+ const family = id.split(":")[0]?.toLowerCase() ?? "";
963
+ for (const entry of FAMILY_DEFAULTS) if (family.startsWith(entry.prefix) && levels.includes(entry.effort)) return entry.effort;
964
+ }
965
+ /** Normalize boolean switches and level ladders, then apply the known family default. */
966
+ function normalizeThinking(id, levels, toggle, fallback) {
967
+ const reasoning = normalizeEfforts(levels);
968
+ if (reasoning.length === 0 && toggle) reasoning.push("off", "high");
969
+ const candidate = fallback !== void 0 && reasoning.includes(fallback) ? fallback : defaultEffortFor(id, reasoning) ?? (toggle && reasoning.includes("high") ? "high" : void 0);
970
+ return {
971
+ reasoning,
972
+ ...candidate === void 0 ? {} : { defaultReasoning: candidate }
973
+ };
775
974
  }
776
975
  // src/catalog.ts
777
976
  const CATALOG_TTL_MS = 6e5;
@@ -836,6 +1035,7 @@ function metaFromLive(entry, caps) {
836
1035
  const toggle = (caps?.has("thinking") === true || levels.length > 0) && levels.length === 0;
837
1036
  const modalities = strings(entry.input_modalities);
838
1037
  const image = modalities.includes("image") || base?.input.includes("image") === true && modalities.length === 0;
1038
+ const normalized = normalizeThinking(id, levels.length > 0 ? levels : base?.reasoning ?? [], toggle || base?.reasoningToggle === true, str$1(entry.default_reasoning_level) ?? base?.defaultReasoning);
839
1039
  return {
840
1040
  id,
841
1041
  name: str$1(entry.display_name) ?? id,
@@ -843,12 +1043,36 @@ function metaFromLive(entry, caps) {
843
1043
  contextWindow: num$1(entry.context_window) ?? num$1(entry.max_context_window) ?? base?.contextWindow ?? 131072,
844
1044
  maxTokens: base?.maxTokens ?? 32768,
845
1045
  input: image ? ["text", "image"] : ["text"],
846
- reasoning: levels.length > 0 ? levels : toggle ? ["off", "high"] : base?.reasoning ?? [],
847
- reasoningToggle: toggle || base?.reasoningToggle === true && levels.length === 0,
1046
+ reasoning: normalized.reasoning,
1047
+ reasoningToggle: toggle || base?.reasoningToggle === true,
848
1048
  tools: caps?.has("tools") === true || base?.tools === true || base === void 0,
849
1049
  free: base?.free === true,
850
1050
  available: entry.supported_in_api !== false && entry.visibility !== "hide",
851
- ...str$1(entry.default_reasoning_level) === void 0 ? base?.defaultReasoning === void 0 ? {} : { defaultReasoning: base.defaultReasoning } : { defaultReasoning: str$1(entry.default_reasoning_level) },
1051
+ ...normalized.defaultReasoning === void 0 ? {} : { defaultReasoning: normalized.defaultReasoning },
1052
+ cost: base?.cost ?? {
1053
+ input: 0,
1054
+ output: 0,
1055
+ cacheRead: 0
1056
+ }
1057
+ };
1058
+ }
1059
+ function metaFromProbe(id, probe) {
1060
+ if (probe === void 0) return void 0;
1061
+ const base = GENERATED_MODELS[id];
1062
+ const normalized = normalizeThinking(id, probe.reasoning.length > 0 ? probe.reasoning : base?.reasoning ?? [], probe.reasoningToggle || base?.reasoningToggle === true, base?.defaultReasoning);
1063
+ return {
1064
+ id,
1065
+ name: base?.name ?? id,
1066
+ transport: base?.transport ?? "native",
1067
+ contextWindow: probe.contextWindow ?? base?.contextWindow ?? 131072,
1068
+ maxTokens: base?.maxTokens ?? 32768,
1069
+ input: [...probe.input],
1070
+ reasoning: normalized.reasoning,
1071
+ reasoningToggle: probe.reasoningToggle || base?.reasoningToggle === true,
1072
+ tools: probe.tools || base?.tools === true,
1073
+ free: base?.free === true,
1074
+ available: probe.available,
1075
+ ...normalized.defaultReasoning === void 0 ? {} : { defaultReasoning: normalized.defaultReasoning },
852
1076
  cost: base?.cost ?? {
853
1077
  input: 0,
854
1078
  output: 0,
@@ -856,6 +1080,13 @@ function metaFromLive(entry, caps) {
856
1080
  }
857
1081
  };
858
1082
  }
1083
+ async function adoptProbes(config, ids, base, fetchImpl) {
1084
+ const probes = await Promise.all(ids.slice(0, 64).map(async (id) => ({
1085
+ id,
1086
+ meta: metaFromProbe(id, await probeModel(config, id, fetchImpl))
1087
+ })));
1088
+ for (const { id, meta } of probes) if (meta !== void 0) liveMeta.set(liveKey(base, id), meta);
1089
+ }
859
1090
  /** Replace the in-memory live metadata for one API base with one successful response. */
860
1091
  function adoptLive(models, tags, base) {
861
1092
  const caps = capabilitiesByModel(tags);
@@ -923,13 +1154,16 @@ async function refreshNow(config, fetchImpl, key, keepOnFailure, epoch) {
923
1154
  }
924
1155
  }
925
1156
  const fromData = Array.isArray(payload.data) ? payload.data.map((entry) => str$1(entry.id)).filter((id) => id !== void 0) : [];
926
- if (fromData.length > 0) {
1157
+ const fromTags = ((tagsResult.status === "fulfilled" ? tagsResult.value : void 0)?.models ?? []).map((entry) => str$1(entry.name) ?? str$1(entry.model)).filter((id) => id !== void 0);
1158
+ const ids = fromData.length > 0 ? fromData : fromTags;
1159
+ if (ids.length > 0) {
927
1160
  for (const liveKey of liveMeta.keys()) if (liveKey.startsWith(`${key}|`)) liveMeta.delete(liveKey);
1161
+ if (isLocalEndpoint(config)) await adoptProbes(config, ids, key, fetchImpl);
928
1162
  cache$1.set(key, {
929
1163
  at: Date.now(),
930
- ids: fromData
1164
+ ids
931
1165
  });
932
- return fromData;
1166
+ return ids;
933
1167
  }
934
1168
  }
935
1169
  if (keepOnFailure) {
@@ -955,6 +1189,7 @@ function clearCatalogCache() {
955
1189
  cache$1.clear();
956
1190
  inflight.clear();
957
1191
  liveMeta.clear();
1192
+ clearModelProbeCache();
958
1193
  }
959
1194
  function baseMeta(id, base) {
960
1195
  const meta = liveMeta.get(liveKey(base, id)) ?? GENERATED_MODELS[id];
@@ -1392,10 +1627,19 @@ function buildNativeBody(options, model, config, images) {
1392
1627
  const tools = buildTools(options.tools);
1393
1628
  const maxTokens = options.maxTokens ?? effectiveMaxTokens(model, config);
1394
1629
  const think = nativeThink(options.reasoningEffort === void 0 ? void 0 : String(options.reasoningEffort), model);
1630
+ const temperature = options.temperature ?? config.temperature;
1631
+ const stop = options.stop ?? config.stop;
1395
1632
  const nativeOptions = {
1396
- ...options.temperature === void 0 ? {} : { temperature: options.temperature },
1633
+ ...temperature === void 0 ? {} : { temperature },
1634
+ ...config.topK === void 0 ? {} : { top_k: config.topK },
1635
+ ...config.topP === void 0 ? {} : { top_p: config.topP },
1636
+ ...config.minP === void 0 ? {} : { min_p: config.minP },
1637
+ ...config.seed === void 0 ? {} : { seed: config.seed },
1638
+ ...config.repeatPenalty === void 0 ? {} : { repeat_penalty: config.repeatPenalty },
1639
+ ...config.repeatLastN === void 0 ? {} : { repeat_last_n: config.repeatLastN },
1640
+ ...config.numCtx === void 0 ? {} : { num_ctx: config.numCtx },
1397
1641
  ...maxTokens === void 0 || maxTokens <= 0 ? {} : { num_predict: maxTokens },
1398
- ...options.stop === void 0 || options.stop.length === 0 ? {} : { stop: options.stop }
1642
+ ...stop === void 0 || stop.length === 0 ? {} : { stop }
1399
1643
  };
1400
1644
  return {
1401
1645
  model: options.model,
@@ -1403,6 +1647,7 @@ function buildNativeBody(options, model, config, images) {
1403
1647
  stream: true,
1404
1648
  ...tools === void 0 ? {} : { tools },
1405
1649
  ...think === void 0 ? {} : { think },
1650
+ ...config.keepAlive === void 0 ? {} : { keep_alive: config.keepAlive },
1406
1651
  ...Object.keys(nativeOptions).length === 0 ? {} : { options: nativeOptions }
1407
1652
  };
1408
1653
  }
@@ -1493,6 +1738,8 @@ function buildChatBody(options, model, config, images) {
1493
1738
  const maxTokens = options.maxTokens ?? effectiveMaxTokens(model, config);
1494
1739
  const effort = options.reasoningEffort === void 0 ? void 0 : String(options.reasoningEffort);
1495
1740
  const reasoning = effort === void 0 || effort === "" ? void 0 : effort === "off" ? "none" : effort;
1741
+ const temperature = options.temperature ?? config.temperature;
1742
+ const stop = options.stop ?? config.stop;
1496
1743
  return {
1497
1744
  model: options.model,
1498
1745
  messages: buildChatMessages(options, images),
@@ -1502,10 +1749,12 @@ function buildChatBody(options, model, config, images) {
1502
1749
  tools,
1503
1750
  tool_choice: "auto"
1504
1751
  },
1505
- ...options.temperature === void 0 ? {} : { temperature: options.temperature },
1752
+ ...temperature === void 0 ? {} : { temperature },
1753
+ ...config.topP === void 0 ? {} : { top_p: config.topP },
1754
+ ...config.seed === void 0 ? {} : { seed: config.seed },
1506
1755
  ...maxTokens === void 0 || maxTokens <= 0 ? {} : { max_tokens: maxTokens },
1507
1756
  ...reasoning === void 0 ? {} : { reasoning_effort: reasoning },
1508
- ...options.stop === void 0 || options.stop.length === 0 ? {} : { stop: options.stop }
1757
+ ...stop === void 0 || stop.length === 0 ? {} : { stop }
1509
1758
  };
1510
1759
  }
1511
1760
  // src/provider-ids.ts
@@ -2581,7 +2830,7 @@ var OllamaCloudAdapter = class extends LlmAdapter {
2581
2830
  async send(url, key, model, body, config, signal) {
2582
2831
  const headers = {
2583
2832
  ...attributionHeaders(),
2584
- authorization: `Bearer ${key}`,
2833
+ ...key === "" ? {} : { authorization: `Bearer ${key}` },
2585
2834
  "content-type": "application/json",
2586
2835
  accept: model.transport === "openai-chat" ? "text/event-stream" : "application/x-ndjson",
2587
2836
  "user-agent": config.userAgent
@@ -3047,6 +3296,12 @@ async function resolveSlot(ctx, slot) {
3047
3296
  function createKeyPool(ctx, config) {
3048
3297
  const all = async () => {
3049
3298
  const current = config();
3299
+ const hasConfiguredKey = current.apiKey !== void 0 || current.accounts.some((account) => typeof account.apiKey === "string" && account.apiKey.trim() !== "" || typeof account.apiKeyEnv === "string" && account.apiKeyEnv.trim() !== "");
3300
+ if (current.apiKeyEnv === "" && !hasConfiguredKey) return [{
3301
+ key: "",
3302
+ source: "none",
3303
+ label: "default"
3304
+ }];
3050
3305
  const pinned = current.activeAccount.trim();
3051
3306
  const slots = slotsOf(current);
3052
3307
  if (pinned !== "") {
@@ -3693,6 +3948,16 @@ function publicConfig(config) {
3693
3948
  contextWindow: config.contextWindow ?? null,
3694
3949
  contextWindowOverrides: config.contextWindowOverrides,
3695
3950
  maxOutputTokens: config.maxOutputTokens ?? null,
3951
+ temperature: config.temperature ?? null,
3952
+ topK: config.topK ?? null,
3953
+ topP: config.topP ?? null,
3954
+ minP: config.minP ?? null,
3955
+ seed: config.seed ?? null,
3956
+ repeatPenalty: config.repeatPenalty ?? null,
3957
+ repeatLastN: config.repeatLastN ?? null,
3958
+ numCtx: config.numCtx ?? null,
3959
+ keepAlive: config.keepAlive ?? "",
3960
+ stop: config.stop ?? [],
3696
3961
  accounts: config.accounts.map((account) => ({
3697
3962
  label: typeof account.label === "string" ? account.label : "",
3698
3963
  apiKeyEnv: typeof account.apiKeyEnv === "string" ? account.apiKeyEnv : ""
@@ -3999,6 +4264,7 @@ function configOps(patch) {
3999
4264
  };
4000
4265
  if ("apiKeyEnv" in patch) {
4001
4266
  if (patch.apiKeyEnv === null || patch.apiKeyEnv === void 0) set("apiKeyEnv", void 0);
4267
+ else if (typeof patch.apiKeyEnv === "string" && patch.apiKeyEnv.trim() === "") set("apiKeyEnv", "");
4002
4268
  else {
4003
4269
  const ref = normalizeCredentialRef(patch.apiKeyEnv);
4004
4270
  if (ref === void 0) throw new Error("invalid credential reference");
@@ -4020,6 +4286,16 @@ function configOps(patch) {
4020
4286
  if ("contextWindow" in patch) set("contextWindow", patch.contextWindow);
4021
4287
  if ("contextWindowOverrides" in patch) set("contextWindowOverrides", cleanNumberRecord(patch.contextWindowOverrides));
4022
4288
  if ("maxOutputTokens" in patch) set("maxOutputTokens", patch.maxOutputTokens);
4289
+ if ("temperature" in patch) set("temperature", patch.temperature);
4290
+ if ("topK" in patch) set("topK", patch.topK);
4291
+ if ("topP" in patch) set("topP", patch.topP);
4292
+ if ("minP" in patch) set("minP", patch.minP);
4293
+ if ("seed" in patch) set("seed", patch.seed);
4294
+ if ("repeatPenalty" in patch) set("repeatPenalty", patch.repeatPenalty);
4295
+ if ("repeatLastN" in patch) set("repeatLastN", patch.repeatLastN);
4296
+ if ("numCtx" in patch) set("numCtx", patch.numCtx);
4297
+ if ("keepAlive" in patch) set("keepAlive", patch.keepAlive);
4298
+ if ("stop" in patch) set("stop", cleanStringArray(patch.stop));
4023
4299
  if ("accounts" in patch) {
4024
4300
  const accounts = cleanAccounts(patch.accounts);
4025
4301
  for (const account of accounts) if (account.apiKeyEnv !== "" && !isCredentialRef(account.apiKeyEnv)) throw new Error("invalid credential reference");
@@ -4771,6 +5047,7 @@ function createUsageReader(config, resolveKey, fetchImpl = fetch) {
4771
5047
  });
4772
5048
  };
4773
5049
  const readBalance = async (key) => {
5050
+ if (key === "") return void 0;
4774
5051
  const current = config();
4775
5052
  try {
4776
5053
  const response = await request(current.balanceUrl, key);
@@ -4781,6 +5058,7 @@ function createUsageReader(config, resolveKey, fetchImpl = fetch) {
4781
5058
  }
4782
5059
  };
4783
5060
  const readUsage = async (key, range, epoch) => {
5061
+ if (key === "") return void 0;
4784
5062
  const current = config();
4785
5063
  const cacheKey = `${fingerprint(key)}:${range}`;
4786
5064
  const hit = cached(cacheKey);
@@ -5101,6 +5379,6 @@ function apply(ctx, config) {
5101
5379
  }), "dsh-ollama-cloud-provider: plugin updater");
5102
5380
  });
5103
5381
  }
5104
- export { Config, OllamaCloudAdapter, PROVIDER, apply, applyCommands, applyRoutes, classifyResponseError, clearCatalogCache, createKeyPool, createKeyResolver, createUsageReader, describeModel, effectiveModel, inject, name, normalizeBalance, normalizeUsage, resolveConfig };
5382
+ export { Config, OllamaCloudAdapter, PROVIDER, apply, applyCommands, applyRoutes, classifyResponseError, clearCatalogCache, clearModelProbeCache, createKeyPool, createKeyResolver, createUsageReader, describeModel, effectiveModel, inject, modelProbeCacheInfo, name, normalizeBalance, normalizeUsage, probeModel, resolveConfig };
5105
5383
 
5106
5384
  //# sourceMappingURL=index.js.map
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@goodandready/dsh-ollama-cloud-provider",
3
- "version": "0.1.17",
3
+ "version": "0.1.19",
4
4
  "description": "Ollama Cloud LLM provider plugin for DeepSeek Harness (dsh).",
5
5
  "license": "MIT",
6
6
  "type": "module",
@@ -73,7 +73,7 @@
73
73
  "build": "tsdown",
74
74
  "prepare": "tsdown 1>&2",
75
75
  "typecheck": "tsc --noEmit",
76
- "test": "node --import tsx --test tests/core.test.ts tests/usage.test.ts tests/adapter.test.ts tests/routes.test.ts tests/native.test.ts tests/openai-chat.test.ts tests/commands.test.ts tests/tui-settings.test.ts tests/analytics.test.ts tests/apply.test.ts tests/account-rules.test.ts tests/request-log.test.ts tests/image-budget.test.ts tests/tool-schema.test.ts tests/tools.test.ts tests/plugin-updater.test.ts tests/compat.test.ts tests/status.test.ts tests/settings-draft.test.ts tests/update-state.test.ts tests/ui-render.test.tsx tests/update-card-view.test.tsx tests/web.test.ts tests/mock-ollama.test.ts",
76
+ "test": "node --import tsx --test tests/core.test.ts tests/usage.test.ts tests/adapter.test.ts tests/routes.test.ts tests/native.test.ts tests/openai-chat.test.ts tests/commands.test.ts tests/tui-settings.test.ts tests/analytics.test.ts tests/apply.test.ts tests/account-rules.test.ts tests/request-log.test.ts tests/image-budget.test.ts tests/tool-schema.test.ts tests/tools.test.ts tests/plugin-updater.test.ts tests/compat.test.ts tests/status.test.ts tests/settings-draft.test.ts tests/update-state.test.ts tests/ui-render.test.tsx tests/update-card-view.test.tsx tests/web.test.ts tests/mock-ollama.test.ts tests/local.test.ts tests/reasoning.test.ts",
77
77
  "check": "npm run typecheck && npm test && npm run build",
78
78
  "release": "npm run check && npm pack",
79
79
  "sync:models": "node scripts/sync-models.mjs"