codeep 2.16.0 → 2.18.0

package/README.md

@@ -9,7 +9,7 @@
 </p>
 
 <p align="center">
-  <img src="Codeep.gif" alt="Codeep demo" width="700">
+  <img src="promo/sources/tui-current.png" alt="Codeep terminal UI with agent timeline, changed files, checks, and resource estimates" width="1000">
 </p>
 
 <p align="center">
@@ -23,6 +23,16 @@
   <a href="https://github.com/VladoIvankovic/Codeep"><img src="https://img.shields.io/github/stars/VladoIvankovic/Codeep?style=social" alt="GitHub stars"></a>
 </p>
 
+## One agent, every surface
+
+Codeep keeps the same project context, provider configuration, sessions, and
+agent workflow wherever you work:
+
+- **[Terminal / TUI](https://codeep.dev/docs/installation)** — the primary CLI experience, with Plan → Read → Edit → Verify → Summary, live file changes, checks, token cost, and estimated energy and cooling-water impact.
+- **[macOS app](https://codeep.dev/docs/mac)** — a native SwiftUI workbench with project conversations and Run Inspector.
+- **[VS Code extension](https://github.com/VladoIvankovic/Codeep-vscode)** — the same agent over ACP, directly inside the editor. Also available on the [VS Code Marketplace](https://marketplace.visualstudio.com/items?itemName=VladoIvankovic.codeep).
+- **[Web dashboard](https://codeep.dev/dashboard)** — synced sessions, usage, costs, tasks, and project activity.
+
 ## Upgrading
 
 Running `npm install -g codeep@latest` picks up the new version — config,
@@ -1029,11 +1039,11 @@ After installation, `codeep` is available globally in your terminal. Simply run
 > /profile save fast
 
 > /provider      # switch to openai
-> /model gpt-4.1
+> /model gpt-5.6-sol
 > /profile save work
 
 > /model fast    # instantly switch to z.ai / glm-5.2
-> /model work    # instantly switch to openai / gpt-4.1
+> /model work    # instantly switch to openai / gpt-5.6-sol
 ```
 
 ### Session Management
@@ -1265,6 +1275,17 @@ With write access enabled:
 | `GOOGLE_API_KEY` | Google AI (Gemini) API key |
 | `MINIMAX_API_KEY` | MiniMax (international) API key |
 | `MINIMAX_CN_API_KEY` | MiniMax China API key |
+| `KIMI_CODE_API_KEY` | Kimi Code subscription key |
+| `MOONSHOT_API_KEY` | Kimi international pay-per-use key |
+| `MOONSHOT_CN_API_KEY` | Kimi China pay-per-use key |
+| `XAI_API_KEY` | Grok (xAI) API key |
+| `BAILIAN_CODING_PLAN_API_KEY` | Qwen international Coding Plan key |
+| `BAILIAN_TOKEN_PLAN_API_KEY` | Qwen international Token Plan key |
+| `DASHSCOPE_API_KEY` | Qwen international pay-per-use key |
+| `BAILIAN_CODING_PLAN_CN_API_KEY` | Qwen China Coding Plan key |
+| `DASHSCOPE_CN_API_KEY` | Qwen China pay-per-use key |
+| `MODELSCOPE_API_KEY` | ModelScope token for its live model catalog |
+| `OPENROUTER_API_KEY` | OpenRouter API key |
 
 ### Settings (`/settings`)
 
@@ -1366,6 +1387,10 @@ Contributions are welcome! Before opening a PR, please read the
 how to propose a change. For provider integrations specifically, the guide
 walks through the files you need to touch.
 
+The active-product scope, model source index, replacement rules, telemetry
+contract, and recurring review checklist live in the
+[**model maintenance policy**](./docs/MODEL_MAINTENANCE.md).
+
 Quick version: `npm install` → `npm test` → `npx tsc --noEmit` clean → open a
 PR against `main`. To report a security issue, see
 [**Security policy**](./SECURITY.md).
@@ -1473,7 +1498,7 @@ The chat sidebar now surfaces two extra ACP signals that previously only the TUI
 ### Editor-native actions (new in 2.2)
 
 - **Code Actions (lightbulb)** — select code and press `Ctrl+.` for **Explain**, **Improve / refactor**, **Add tests**, and **Add doc comment**. On a line with an error/warning, a **Fix this problem** quick-fix sends the diagnostic plus the code to Codeep. Everything routes through the chat, so the full agent (file edits via the diff preview, MCP tools) is available.
-- **Model picker in the status bar** — click `Codeep · <model>` (or run **Codeep: Select Provider & Model**) to switch provider + model from a quick-pick. Providers with open-ended catalogs (OpenRouter, Ollama, custom endpoints) let you type a model id.
+- **Model picker in the status bar** — click `Codeep · <model>` (or run **Codeep: Select Provider & Model**) to switch provider + model from a quick-pick. Providers with open-ended catalogs (OpenRouter, ModelScope, Ollama, custom endpoints) let you type a model id.
 - **Self-hosted endpoints from settings** — point the extension at vLLM / LiteLLM / LM Studio / text-generation-webui with `codeep.baseUrl` (e.g. `http://localhost:8000/v1`), plus `codeep.provider` (`custom` or `openai`) and `codeep.model`. The `codeep.provider` / `codeep.model` settings are applied on every connect, so they stay authoritative.
 - **Get Started walkthrough** — a native VS Code walkthrough (Help → Get Started) covering CLI install, opening the chat, editor actions, and choosing a model.
 

package/dist/acp/serverHandlers.js

@@ -224,7 +224,7 @@ export function buildProviderList() {
         // Model metadata so ACP clients (e.g. the VS Code model picker) can
         // offer a provider → model selector without hardcoding a catalog.
         // `dynamicModels` flags providers whose model list is open-ended
-        // (OpenRouter, Ollama, custom endpoints) — clients should let the
+        // (OpenRouter, ModelScope, Ollama, custom endpoints) — clients should let the
         // user type a model id rather than only pick from `models`.
         models: p.models.map((m) => ({ id: m.id, name: m.name })),
         defaultModel: p.defaultModel,

package/dist/api/index.js

@@ -343,7 +343,13 @@ async function chatOpenAI(message, history, model, apiKey, onChunk, abortSignal)
     let baseUrl = resolveBaseUrl(providerId, 'openai');
     const authHeader = getProviderAuthHeader(providerId, 'openai');
     const useCompletionTokens = usesMaxCompletionTokens(providerId);
-    const omitTemperature = requiresDefaultTemperature(providerId);
+    // Two independent reasons to leave temperature out, and both must be checked
+    // here: the provider-level flag (endpoints that only accept the default) and
+    // the per-model one (generations that removed the sampling params outright,
+    // e.g. gemini-3.7-flash). Testing only the provider flag let a rejecting
+    // model through, because every model this rule covers is served over the
+    // OpenAI-compatible path, not the Anthropic one.
+    const omitTemperature = requiresDefaultTemperature(providerId) || modelRejectsSamplingParams(model);
     if (!baseUrl) {
         throw new Error(`Provider ${providerId} does not support OpenAI protocol`);
     }

package/dist/config/index.js

@@ -2,7 +2,7 @@ import Conf from 'conf';
 import { existsSync, mkdirSync, readdirSync, readFileSync, writeFileSync, unlinkSync, statSync } from 'fs';
 import { join, dirname } from 'path';
 import { randomUUID } from 'crypto';
-import { PROVIDERS, getProvider, getProviderBaseUrl } from './providers.js';
+import { PROVIDERS, getProvider, getProviderBaseUrl, replacementModelFor } from './providers.js';
 import { logSession } from '../utils/logger.js';
 import { createSecureStorage } from '../utils/keychain.js';
 // We'll initialize GLOBAL_SESSIONS_DIR after config is created (to use config.path)
@@ -255,8 +255,9 @@ if (config.get('agentMode') === 'auto') {
 // affected sliders were effectively lies). Each migration now runs exactly
 // once per config, recorded via `migrationVersion`; after that, whatever the
 // user sets sticks. Bump MIGRATION_VERSION when adding a new one.
-const MIGRATION_VERSION = 1;
-if ((config.get('migrationVersion') ?? 0) < 1) {
+const MIGRATION_VERSION = 4;
+const currentMigrationVersion = config.get('migrationVersion') ?? 0;
+if (currentMigrationVersion < 1) {
     // Migrate the old runaway default (10000 iterations) down to a sane
     // ceiling, and old conservative defaults up to the current ones.
     if (config.get('agentMaxIterations') >= 10000) {
@@ -277,6 +278,18 @@ if ((config.get('migrationVersion') ?? 0) < 1) {
     if (config.get('rateLimitCommands') <= 100) {
         config.set('rateLimitCommands', 10000);
     }
+}
+if (currentMigrationVersion < 4) {
+    // Vendor aliases below were removed from Codeep's curated catalogue after
+    // their replacements became available. Migrate only exact known aliases:
+    // OpenRouter/Ollama/custom model ids remain user-controlled.
+    const provider = config.get('provider');
+    const model = config.get('model');
+    const replacement = replacementModelFor(provider, model);
+    if (replacement)
+        config.set('model', replacement);
+}
+if (currentMigrationVersion < MIGRATION_VERSION) {
     config.set('migrationVersion', MIGRATION_VERSION);
 }
 // Global sessions directory - use same directory as conf package for cross-platform consistency
@@ -1216,7 +1229,10 @@ export function loadProfile(name) {
 }
 export function applyProfile(profile) {
     config.set('provider', profile.provider);
-    config.set('model', profile.model);
+    // A profile saved before a vendor retired a model would otherwise restore an
+    // id the picker no longer offers. Normalize it exactly like the startup
+    // migration does — dynamic OpenRouter/Ollama/custom ids are never rewritten.
+    config.set('model', replacementModelFor(profile.provider, profile.model) ?? profile.model);
     config.set('protocol', profile.protocol);
     config.set('temperature', profile.temperature);
     config.set('maxTokens', profile.maxTokens);

package/dist/config/providers.d.ts

@@ -34,6 +34,10 @@ export interface ProviderConfig {
     subscribeUrl?: string;
     noApiKey?: boolean;
     dynamicModels?: boolean;
+    /** Billed as a flat subscription (or free) — token counts are real, per-token
+     *  cost is not. Cost surfaces must say "included in plan" instead of pricing
+     *  these tokens at the provider's pay-per-use rates. */
+    flatFee?: boolean;
     groupLabel?: string;
     hint?: string;
     mcpEndpoints?: {
@@ -45,6 +49,7 @@ export interface ProviderConfig {
 export declare const PROVIDERS: Record<string, ProviderConfig>;
 export type ProviderId = keyof typeof PROVIDERS;
 export declare function getProvider(id: string): ProviderConfig | null;
+export declare function replacementModelFor(providerId: string, modelId: string): string | undefined;
 export declare function getProviderList(): {
     id: string;
     name: string;
@@ -59,6 +64,11 @@ export declare function getProviderModels(providerId: string): {
 }[];
 export declare function isNoApiKeyProvider(providerId: string): boolean;
 export declare function isDynamicModelsProvider(providerId: string): boolean;
+/**
+ * Returns true if the provider bills a flat subscription (or is free), so any
+ * per-token dollar figure we compute for it is invented — see `flatFee`.
+ */
+export declare function isFlatFeeProvider(providerId: string): boolean;
 export declare function getProviderBaseUrl(providerId: string, protocol: 'openai' | 'anthropic'): string | null;
 export declare function getProviderAuthHeader(providerId: string, protocol: 'openai' | 'anthropic'): 'Bearer' | 'x-api-key';
 export declare function getProviderMcpEndpoints(providerId: string): ProviderConfig['mcpEndpoints'] | null;
@@ -133,8 +143,8 @@ export declare function availableReasoningTiers(providerId: string, model: strin
 /**
  * Map a (possibly out-of-range) tier to the tier this model actually distinguishes,
  * for display — the chip + the checked menu row. The effort setting is global, so
- * a tier picked on Opus ('low') may not exist on GLM-5.2; we show the level GLM
- * will really run (its 'low' clamps to 'high'). Picks the available tier whose
+ * a tier picked on Opus ('medium') may not exist on Kimi K3; we show the level
+ * Kimi will really run (its 'medium' clamps to 'high'). Picks the tier whose
  * effective param equals the requested one. 'auto' (or unsupported) → 'auto'.
  */
 export declare function resolveReasoningTier(providerId: string, model: string, tier: ReasoningTier): ReasoningTier;