@aliou/pi-neuralwatt 0.5.2 → 0.6.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@aliou/pi-neuralwatt",
-  "version": "0.5.2",
+  "version": "0.6.0",
   "license": "MIT",
   "type": "module",
   "private": false,
@@ -32,7 +32,7 @@
     "!src/**/*.test.ts"
   ],
   "dependencies": {
-    "@aliou/pi-utils-settings": "^0.15.0",
+    "@aliou/pi-utils-settings": "^0.17.0",
     "@aliou/pi-utils-ui": "^0.4.0"
   },
   "peerDependencies": {
@@ -51,12 +51,10 @@
     "typescript": "^5.9.3",
     "vitest": "^4.0.18"
   },
-  "peerDependenciesMeta": {
-    "@earendil-works/pi-coding-agent": {
-      "optional": true
-    },
-    "@earendil-works/pi-tui": {
-      "optional": true
+  "pnpm": {
+    "overrides": {
+      "@earendil-works/pi-ai": "$@earendil-works/pi-coding-agent",
+      "@earendil-works/pi-tui": "$@earendil-works/pi-coding-agent"
     }
   },
   "scripts": {
@@ -68,8 +66,18 @@
     "gen:schema": "ts-json-schema-generator --path src/config.ts --type NeuralwattConfig --no-type-check -o schema.json",
     "check:schema": "ts-json-schema-generator --path src/config.ts --type NeuralwattConfig --no-type-check -o /tmp/schema-check.json && diff -q schema.json /tmp/schema-check.json",
     "check:lockfile": "pnpm install --frozen-lockfile --ignore-scripts",
+    "prepare": "[ -d .git ] && husky || true",
     "changeset": "changeset",
     "version": "changeset version",
     "release": "pnpm changeset publish"
+  },
+  "packageManager": "pnpm@10.26.1",
+  "peerDependenciesMeta": {
+    "@earendil-works/pi-coding-agent": {
+      "optional": true
+    },
+    "@earendil-works/pi-tui": {
+      "optional": true
+    }
   }
-}
+}

package/schema.json

@@ -5,6 +5,10 @@
     "NeuralwattConfig": {
       "additionalProperties": false,
       "properties": {
+        "includeLegacyModelIds": {
+          "description": "Include legacy Neuralwatt model IDs in the model picker.",
+          "type": "boolean"
+        },
         "quotaCommand": {
           "description": "Show the quota command (/neuralwatt:quota).",
           "type": "boolean"

package/src/config.ts

@@ -28,24 +28,38 @@ export interface NeuralwattConfig {
   quotaWarnings?: boolean;
   /** Show usage in the sub-bar / status bar. */
   subBarIntegration?: boolean;
+  /** Include legacy Neuralwatt model IDs in the model picker. */
+  includeLegacyModelIds?: boolean;
 }
 
 export interface ResolvedNeuralwattConfig {
   quotaCommand: boolean;
   quotaWarnings: boolean;
   subBarIntegration: boolean;
+  includeLegacyModelIds: boolean;
 }
 
 const DEFAULTS: ResolvedNeuralwattConfig = {
   quotaCommand: true,
   quotaWarnings: true,
   subBarIntegration: true,
+  includeLegacyModelIds: false,
 };
 
 export const configLoader = new ConfigLoader<
   NeuralwattConfig,
   ResolvedNeuralwattConfig
->("neuralwatt", DEFAULTS);
+>("neuralwatt", DEFAULTS, {
+  migrations: [
+    {
+      name: "disable-legacy-model-ids-by-default",
+      shouldRun: (config) => config.includeLegacyModelIds === undefined,
+      message:
+        "[neuralwatt] legacy model IDs (ids including the provider and the quantization) are disabled by default. You can enable them with /neuralwatt:settings.",
+      run: (config) => ({ ...config, includeLegacyModelIds: false }),
+    },
+  ],
+});
 
 export const NEURALWATT_CONFIG_UPDATED_EVENT =
   "neuralwatt:config:updated" as const;
@@ -128,6 +142,23 @@ export function registerNeuralwattSettings(
             ),
           ],
         },
+        {
+          label: "Other settings",
+          items: [
+            {
+              id: "includeLegacyModelIds",
+              label: "Legacy model IDs",
+              description:
+                "Include deprecated Neuralwatt model IDs as aliases in the model picker",
+              currentValue:
+                (tabConfig?.includeLegacyModelIds ??
+                resolved.includeLegacyModelIds)
+                  ? "include"
+                  : "ignore",
+              values: ["include", "ignore"],
+            },
+          ],
+        },
       ];
     },
     onSettingChange: (id, newValue, config) => {
@@ -135,6 +166,10 @@ export function registerNeuralwattSettings(
         return null;
       }
 
+      if (id === "includeLegacyModelIds") {
+        return { ...config, includeLegacyModelIds: newValue === "include" };
+      }
+
       const enabled = newValue === "enabled";
       switch (id) {
         case "quotaCommand":

package/src/extensions/provider/index.ts

@@ -2,6 +2,7 @@ import type { ExtensionAPI } from "@earendil-works/pi-coding-agent";
 import {
   configLoader,
   emitConfigUpdated,
+  NEURALWATT_CONFIG_UPDATED_EVENT,
   NEURALWATT_EXTENSIONS_REGISTER_EVENT,
   NEURALWATT_EXTENSIONS_REQUEST_EVENT,
   type NeuralwattFeatureId,
@@ -16,12 +17,14 @@ import {
 } from "../../types/quota-events";
 import { fetchQuotas } from "../../utils/quotas";
 import { normalizeNeuralwattContextOverflowError } from "./context-overflow";
-import { NEURALWATT_MODELS } from "./models";
+import { getNeuralwattModels } from "./models";
 import { buildQuotasFromHeaders, fetchRequestedQuotas } from "./quota-store";
 
 const HEADER_EMIT_THROTTLE_MS = 5_000;
 
 function registerNeuralwattProvider(pi: ExtensionAPI): void {
+  const { includeLegacyModelIds } = configLoader.getConfig();
+
   pi.registerProvider("neuralwatt", {
     baseUrl: "https://api.neuralwatt.com/v1",
     apiKey: "$NEURALWATT_API_KEY",
@@ -31,7 +34,9 @@ function registerNeuralwattProvider(pi: ExtensionAPI): void {
       Referer: "https://pi.dev",
       "X-Title": "npm:@aliou/pi-neuralwatt",
     },
-    models: NEURALWATT_MODELS,
+    models: getNeuralwattModels({
+      includeLegacyModelIds,
+    }),
   });
 }
 
@@ -47,6 +52,10 @@ export default async function (pi: ExtensionAPI) {
     getLoadedFeatures: () => loadedFeatures,
   });
 
+  pi.events.on(NEURALWATT_CONFIG_UPDATED_EVENT, () => {
+    registerNeuralwattProvider(pi);
+  });
+
   let lastHeaderEmitAt = 0;
   let quotaRequestInFlight = false;
 
@@ -98,6 +107,10 @@ export default async function (pi: ExtensionAPI) {
   });
 
   pi.on("session_start", async (_event, ctx) => {
+    for (const message of configLoader.drainMessages()) {
+      ctx.ui.notify(message, "warning");
+    }
+
     loadedFeatures.clear();
     pi.events.emit(NEURALWATT_EXTENSIONS_REQUEST_EVENT, undefined);
     emitConfigUpdated(pi);

package/src/extensions/provider/models.ts

@@ -5,26 +5,6 @@
 import type { ProviderModelConfig } from "@earendil-works/pi-coding-agent";
 
 export const NEURALWATT_MODELS: ProviderModelConfig[] = [
-  // Devstral Small 2 - Mistral
-  {
-    id: "mistralai/Devstral-Small-2-24B-Instruct-2512",
-    name: "Devstral Small 2",
-    reasoning: false,
-    input: ["text", "image"],
-    cost: {
-      input: 0.12,
-      output: 0.35,
-      cacheRead: 0,
-      cacheWrite: 0,
-    },
-    contextWindow: 262128,
-    maxTokens: 32768,
-    compat: {
-      supportsDeveloperRole: false,
-      maxTokensField: "max_tokens",
-      requiresAssistantAfterToolResult: true,
-    },
-  },
   // GLM-5 Fast - ZhipuAI
   {
     id: "glm-5-fast",
@@ -38,7 +18,7 @@ export const NEURALWATT_MODELS: ProviderModelConfig[] = [
       cacheWrite: 0,
     },
     contextWindow: 202736,
-    maxTokens: 32768,
+    maxTokens: 65536,
     compat: {
       supportsDeveloperRole: false,
       maxTokensField: "max_tokens",
@@ -46,7 +26,7 @@ export const NEURALWATT_MODELS: ProviderModelConfig[] = [
   },
   // GLM-5.1 - ZhipuAI
   {
-    id: "zai-org/GLM-5.1-FP8",
+    id: "glm-5.1",
     name: "GLM-5.1",
     reasoning: true,
     input: ["text"],
@@ -57,7 +37,7 @@ export const NEURALWATT_MODELS: ProviderModelConfig[] = [
       cacheWrite: 0,
     },
     contextWindow: 202736,
-    maxTokens: 32768,
+    maxTokens: 65536,
     thinkingLevelMap: {
       minimal: null,
       low: null,
@@ -84,37 +64,10 @@ export const NEURALWATT_MODELS: ProviderModelConfig[] = [
       cacheWrite: 0,
     },
     contextWindow: 202736,
-    maxTokens: 32768,
-    compat: {
-      supportsDeveloperRole: false,
-      maxTokensField: "max_tokens",
-    },
-  },
-  // GPT-OSS 20B - OpenAI
-  {
-    id: "openai/gpt-oss-20b",
-    name: "GPT-OSS 20B",
-    reasoning: true,
-    input: ["text"],
-    cost: {
-      input: 0.03,
-      output: 0.16,
-      cacheRead: 0,
-      cacheWrite: 0,
-    },
-    contextWindow: 16368,
-    maxTokens: 4096,
-    thinkingLevelMap: {
-      minimal: "low",
-      low: "low",
-      medium: "medium",
-      high: "high",
-      xhigh: null,
-    },
+    maxTokens: 65536,
     compat: {
       supportsDeveloperRole: false,
       maxTokensField: "max_tokens",
-      requiresReasoningContentOnAssistantMessages: true,
     },
   },
   // Kimi K2.5 - MoonshotAI
@@ -165,7 +118,7 @@ export const NEURALWATT_MODELS: ProviderModelConfig[] = [
   },
   // Kimi K2.6 - MoonshotAI
   {
-    id: "moonshotai/Kimi-K2.6",
+    id: "kimi-k2.6",
     name: "Kimi K2.6",
     reasoning: true,
     input: ["text", "image"],
@@ -209,36 +162,9 @@ export const NEURALWATT_MODELS: ProviderModelConfig[] = [
       maxTokensField: "max_tokens",
     },
   },
-  // MiniMax M2.5 - MiniMax
-  {
-    id: "MiniMaxAI/MiniMax-M2.5",
-    name: "MiniMax M2.5",
-    reasoning: true,
-    input: ["text"],
-    cost: {
-      input: 0.35,
-      output: 1.38,
-      cacheRead: 0,
-      cacheWrite: 0,
-    },
-    contextWindow: 196592,
-    maxTokens: 65536,
-    thinkingLevelMap: {
-      minimal: null,
-      low: null,
-      medium: "medium",
-      high: null,
-      xhigh: null,
-    },
-    compat: {
-      supportsDeveloperRole: false,
-      maxTokensField: "max_tokens",
-      requiresReasoningContentOnAssistantMessages: true,
-    },
-  },
   // Qwen3.5 397B - Qwen
   {
-    id: "Qwen/Qwen3.5-397B-A17B-FP8",
+    id: "qwen3.5-397b",
     name: "Qwen3.5 397B",
     reasoning: true,
     input: ["text"],
@@ -284,7 +210,7 @@ export const NEURALWATT_MODELS: ProviderModelConfig[] = [
   },
   // Qwen3.6 35B - Qwen
   {
-    id: "Qwen/Qwen3.6-35B-A3B",
+    id: "qwen3.6-35b",
     name: "Qwen3.6 35B",
     reasoning: true,
     input: ["text", "image"],
@@ -295,7 +221,34 @@ export const NEURALWATT_MODELS: ProviderModelConfig[] = [
       cacheWrite: 0,
     },
     contextWindow: 131056,
-    maxTokens: 32768,
+    maxTokens: 65536,
+    thinkingLevelMap: {
+      minimal: null,
+      low: null,
+      medium: "medium",
+      high: null,
+      xhigh: null,
+    },
+    compat: {
+      supportsDeveloperRole: false,
+      maxTokensField: "max_tokens",
+      requiresReasoningContentOnAssistantMessages: true,
+    },
+  },
+  // Kimi K2.7 Code - MoonshotAI
+  {
+    id: "moonshotai/Kimi-K2.7-Code",
+    name: "Kimi K2.7 Code",
+    reasoning: true,
+    input: ["text", "image"],
+    cost: {
+      input: 0.95,
+      output: 4.0,
+      cacheRead: 0,
+      cacheWrite: 0,
+    },
+    contextWindow: 262128,
+    maxTokens: 65536,
     thinkingLevelMap: {
       minimal: null,
       low: null,
@@ -322,10 +275,46 @@ export const NEURALWATT_MODELS: ProviderModelConfig[] = [
       cacheWrite: 0,
     },
     contextWindow: 131056,
-    maxTokens: 32768,
+    maxTokens: 65536,
     compat: {
       supportsDeveloperRole: false,
       maxTokensField: "max_tokens",
     },
   },
 ];
+
+const LEGACY_MODEL_ALIAS_MAP = {
+  "zai-org/GLM-5.1-FP8": "glm-5.1",
+  "moonshotai/Kimi-K2.6": "kimi-k2.6",
+  "Qwen/Qwen3.5-397B-A17B-FP8": "qwen3.5-397b",
+  "Qwen/Qwen3.6-35B-A3B": "qwen3.6-35b",
+} as const;
+
+export const LEGACY_NEURALWATT_MODEL_IDS = new Set<string>(
+  Object.keys(LEGACY_MODEL_ALIAS_MAP),
+);
+
+const LEGACY_NEURALWATT_MODELS: ProviderModelConfig[] = Object.entries(
+  LEGACY_MODEL_ALIAS_MAP,
+).map(([legacyId, canonicalId]) => {
+  const canonical = NEURALWATT_MODELS.find((model) => model.id === canonicalId);
+
+  if (!canonical) {
+    throw new Error(`Missing canonical model for legacy alias ${legacyId}`);
+  }
+
+  return {
+    ...canonical,
+    id: legacyId,
+    name: `${canonical.name} (legacy ID)`,
+  };
+});
+
+export function getNeuralwattModels(options?: {
+  includeLegacyModelIds?: boolean;
+}): ProviderModelConfig[] {
+  if (options?.includeLegacyModelIds)
+    return [...NEURALWATT_MODELS, ...LEGACY_NEURALWATT_MODELS];
+
+  return NEURALWATT_MODELS;
+}