omni-pi 0.4.0 → 0.6.1

package/CHANGELOG.md

@@ -0,0 +1,42 @@
+# Changelog
+
+## 0.6.1 - 2026-03-29
+
+### Provider management
+
+- renamed bundled provider auth management from `/provider-auth` to `/manage-providers`
+- limited `/model-setup list` to removing individual custom model entries instead of deleting whole custom providers
+- documented the bundled provider list directly in `PROVIDERS.md`
+
+### CI and documentation
+
+- added a unified `npm run verify` gate for local development, CI, and publish checks
+- added docs coverage tests so command docs and the bundled provider list fail CI when they drift from the code
+- added a tag-triggered release workflow that re-verifies the repo, creates a GitHub release, and publishes to npm when credentials are configured
+
+## 0.6.0 - 2026-03-27
+
+### Provider management
+
+- restricted `/model-setup` to custom providers and custom model entries stored in `models.json`
+- added `/provider-auth` to remove stored auth for bundled Pi providers from the UI
+- added whole-provider removal for custom providers, not just single-model removal
+- fixed `/model-setup list` so it only shows custom providers/models instead of the full authenticated runtime catalog
+
+### Provider discovery and refresh
+
+- added startup refresh for authenticated, discoverable custom providers
+- preserved dynamic headers and other existing model metadata when custom providers are edited or rediscovered
+- improved custom-provider onboarding so users can add a provider first and discover models automatically
+- stopped persisting invalid `contextWindow: 0` and `maxTokens: 0` values for discovered providers
+
+### Selector and UX improvements
+
+- aligned Omni-Pi setup selectors with Pi-style searchable selection behavior
+- limited the custom searchable selector to 10 visible rows and only enabled search when more than 10 items are present
+- improved bundled command descriptions and provider-management messaging
+
+### Documentation
+
+- documented the split between custom provider setup and bundled provider auth management in `README.md`
+- added `PROVIDERS.md` with guidance for `/model-setup`, `/provider-auth`, and custom provider discovery behavior

package/CREDITS.md

@@ -9,6 +9,17 @@ Omni-Pi exists because of the Pi ecosystem and the work of earlier authors.
 - [@mariozechner/pi-coding-agent](https://www.npmjs.com/package/@mariozechner/pi-coding-agent) by Mario Zechner
   - Pi coding agent package (direct dependency)
 
+## Bundled third-party extensions
+
+- [pi-web-access](https://www.npmjs.com/package/pi-web-access)
+  - Web search and fetch tools for the agent
+- [pi-interview](https://www.npmjs.com/package/pi-interview)
+  - Guided Q&A interface for user clarification
+- [@juanibiapina/pi-powerbar](https://www.npmjs.com/package/@juanibiapina/pi-powerbar) by Juani Biapina
+  - Powerline-style status bar with extensible segments
+- [@juanibiapina/pi-extension-settings](https://www.npmjs.com/package/@juanibiapina/pi-extension-settings) by Juani Biapina
+  - Settings persistence for Pi extensions
+
 ## Workflow and orchestration inspiration
 
 - [can1357/oh-my-pi](https://github.com/can1357/oh-my-pi)

package/PROVIDERS.md

@@ -0,0 +1,76 @@
+# Provider Setup
+
+Omni-Pi separates bundled providers from custom providers.
+
+## `/model-setup`
+
+`/model-setup` is only for custom providers and custom model entries stored in `models.json`.
+
+Use it when you want to configure:
+
+- your own provider id
+- an API type such as OpenAI-compatible or Anthropic-compatible
+- a custom base URL
+- an API key for that custom provider
+- discovered models or manual model entries
+
+`/model-setup list` only shows custom models from `models.json`, and it removes individual custom model entries only.
+
+## Bundled Providers
+
+Pi's bundled providers are still available through the runtime model registry, but Omni-Pi does not manage them through `/model-setup`.
+
+That means:
+
+- `/model-setup` does not add bundled providers
+- `/model-setup` does not list bundled providers
+- `/model-setup` does not remove bundled providers
+
+If a bundled provider already has valid auth in the Pi runtime, Omni-Pi may use it through normal model selection, but its setup is outside the custom-provider flow documented here.
+
+Use `/manage-providers` to list bundled providers that currently have stored auth and remove that auth when needed.
+
+The bundled-provider list below is expected to stay in sync with the exported provider setup list in `src/model-setup.ts`. The test suite checks that this section matches the code list.
+
+### Bundled provider list
+
+- `anthropic` — API key
+- `openai` — API key
+- `openrouter` — API key
+- `google` — API key
+- `github-copilot` — OAuth
+- `openai-codex` — OAuth
+- `xai` — API key
+- `zai` — API key
+- `azure-openai-responses` — API key
+- `nvidia` — API key
+- `together` — API key
+- `synthetic` — API key
+- `nanogpt` — API key
+- `xiaomi` — API key
+- `moonshot` — API key
+- `venice` — API key
+- `kilo` — API key
+- `gitlab-duo` — API key
+- `qwen-portal` — API key
+- `qianfan` — API key
+- `cloudflare-ai-gateway` — API key
+
+## Custom Provider Discovery
+
+For custom providers that expose a compatible model listing endpoint, Omni-Pi can fetch models for you after you add the provider details and credentials.
+
+On launch, Omni-Pi also refreshes authenticated, discoverable custom providers that are already configured in `models.json`.
+
+## When To Use Which Path
+
+Use `/model-setup` when:
+
+- you are adding a non-bundled provider
+- you need a custom base URL
+- you want to manage your own discovered or manual model list
+
+Do not use `/model-setup` when:
+
+- you are trying to manage a built-in Pi provider
+- you expect bundled providers to appear as removable custom entries

package/README.md

@@ -1,8 +1,6 @@
 # Omni-Pi
 
-Omni-Pi is a Pi package built around one user-facing brain.
-
-The goal is simple: talk to a helpful agent, let it interview you until the request is precise, have it write the spec and task breakdown into `.omni/`, then implement the work in bounded slices with explicit verification and progress notes.
+A batteries-included [Pi](https://github.com/badlogic/pi-mono) package that interviews the user, documents the spec, and implements work in bounded slices.
 
 Requires Node.js 22 or newer.
 
@@ -12,15 +10,13 @@ Requires Node.js 22 or newer.
 
 ## What It Does
 
-- One friendly brain talks to the user.
-- `.omni/` remains the durable memory layer for goals, specs, tasks, checks, progress, and decisions.
-- Work is broken into small, verifiable slices before code changes happen.
-- Verification is explicit and recorded alongside implementation progress.
+- One conversational brain interviews the user until the request is precise.
+- Writes specs, tasks, and progress into `.omni/` as durable project memory.
+- Breaks work into small, verifiable slices and implements them one at a time.
+- Bundles web search, guided interviews, themed UI, a task viewer, a powerbar, custom provider/model management, and automatic updates out of the box.
 
 ## Install
 
-Install the standalone executable:
-
 ```bash
 npm install -g omni-pi
 ```
@@ -32,61 +28,95 @@ cd your-project
 omni
 ```
 
-For local development from this checkout:
+Custom provider setup and bundled provider behavior are documented in [PROVIDERS.md](PROVIDERS.md).
 
-```bash
-git clone https://github.com/EdGy2k/Omni-Pi.git
-cd Omni-Pi
-npm install
-npm run chat
-```
+## Features
 
-## Use
+### Bundled Extensions
 
-Start Pi in the project you want to work on, with Omni-Pi installed, and describe what you want.
+| Extension | What it does |
+|-----------|-------------|
+| **omni-core** | Brain workflow, themed header, session init, system prompt injection |
+| **omni-providers** | Model provider wiring |
+| **omni-memory** | `.omni/` durable memory bootstrap |
+| **pi-web-access** | Web search and fetch tools for the agent |
+| **pi-interview** | Guided Q&A when the agent needs clarification |
+| **pi-powerbar** | Powerline-style status bar with segments |
+| **pi-extension-settings** | Settings persistence for extensions |
 
-Example:
+### Commands
 
-```text
-Build a small CLI notes app for me. I want add, list, and search commands. Store data locally in JSON. Ask me any questions you need before you start coding.
-```
+| Command | Description |
+|---------|-------------|
+| `/model-setup` | Add custom providers/models or remove custom model entries |
+| `/manage-providers` | Remove stored auth for bundled providers |
+| `/theme` | Switch between color presets (lavender, ember, ocean, mint, rose, gold, arctic, neon, copper, slate) |
+| `/update` | Check for Omni-Pi updates |
 
-Expected behavior:
+### Keyboard Shortcuts
 
-- Omni-Pi interviews first when the request is underspecified.
-- It writes and updates `.omni/PROJECT.md`, `.omni/SPEC.md`, `.omni/TASKS.md`, `.omni/TESTS.md`, and `.omni/STATE.md`.
-- It hides internal implementation machinery instead of talking about planner/worker/expert handoffs.
-- It implements one bounded slice at a time and runs the planned checks.
+| Shortcut | Description |
+|----------|-------------|
+| `Ctrl+Shift+T` | Toggle the task list widget (`.omni/TASKS.md` + `.omni/STATE.md`) |
 
-## Durable Memory
+### Auto-Updater
 
-Omni-Pi keeps its working notes in `.omni/`:
+Omni-Pi checks for new versions on startup (cached, re-checks every 4 hours). When an update is available, it prompts to install and restart. Pi's own update notification is suppressed to avoid duplication.
 
-- `PROJECT.md` captures the problem, users, constraints, and success criteria.
-- `SPEC.md` captures the exact requested behavior and implementation shape.
-- `TASKS.md` breaks work into bounded slices.
-- `TESTS.md` records the checks for the current slice.
-- `STATE.md`, `SESSION-SUMMARY.md`, and `DECISIONS.md` keep progress and rationale durable across sessions.
+## Provider Support
 
-## Development
+`/model-setup` is for custom providers and custom model entries only.
 
-`npm run chat` launches the local `omni` wrapper from this checkout, which in turn starts Pi with this package loaded.
+Use it when you want to configure:
 
-For local verification:
+- a custom provider id
+- an API type and base URL
+- an API key for that custom provider
+- discovered models or manual model entries
 
-```bash
-npm run check
-npm run lint
-npm test
-```
+Use `/manage-providers` to remove stored auth for bundled Pi providers.
+
+See [PROVIDERS.md](PROVIDERS.md) for the current supported-provider list and auth-management split.
+
+## Durable Memory
+
+Omni-Pi keeps its working notes in `.omni/`:
 
-For package verification:
+| File | Purpose |
+|------|---------|
+| `PROJECT.md` | Problem, users, constraints, success criteria |
+| `SPEC.md` | Exact requested behavior and implementation shape |
+| `TASKS.md` | Work broken into bounded slices |
+| `TESTS.md` | Checks for the current slice |
+| `STATE.md` | Current phase, active task, blockers |
+| `SESSION-SUMMARY.md` | Progress notes across sessions |
+| `DECISIONS.md` | Rationale for key choices |
+
+## Development
 
 ```bash
-npm pack
-npm publish --dry-run
+git clone https://github.com/EdGy2k/Omni-Pi.git
+cd Omni-Pi
+npm install
+npm run chat    # launch locally in dev mode
 ```
 
+| Command | Purpose |
+|---------|---------|
+| `npm run chat` | Launch the local `omni` executable |
+| `npm test` | Run the test suite (Vitest) |
+| `npm run check` | TypeScript type-check |
+| `npm run lint` | Biome lint + format check |
+| `npm run verify` | Full local/CI gate: type-check, lint, test, and package dry-run |
+| `npm run format` | Auto-fix lint and formatting |
+| `npm install -g .` | Install globally from local checkout |
+
+## CI/CD
+
+- Pull requests and pushes to `main` run `npm run verify`.
+- The docs are part of the test contract, including a sync check between `PROVIDERS.md` and the bundled-provider setup list in code.
+- Pushing a `v*` tag runs the release workflow, verifies the repo again, creates a GitHub release, and publishes to npm when `NPM_TOKEN` is configured.
+
 ## Attribution
 
 Omni-Pi builds on the Pi ecosystem. See [CREDITS.md](CREDITS.md).

package/extensions/omni-core/index.ts

@@ -7,20 +7,26 @@ import {
 import { renderHeader } from "../../src/header.js";
 import { registerModelCommand } from "../../src/model-command.js";
 import { registerOmniMessageRenderer } from "../../src/pi.js";
+import { registerProviderAuthCommand } from "../../src/provider-auth-command.js";
 import { createOmniTheme } from "../../src/theme.js";
 import { registerThemeCommand } from "../../src/theme-command.js";
+import { registerTodoShortcut } from "../../src/todo-shortcut.js";
+import { registerUpdater } from "../../src/updater.js";
 
 export default function omniCoreExtension(api: ExtensionAPI): void {
   registerOmniMessageRenderer(api);
   registerModelCommand(api);
+  registerProviderAuthCommand(api);
   registerThemeCommand(api);
+  registerTodoShortcut(api);
+  registerUpdater(api);
 
   api.on("session_start", async (_event, ctx) => {
     await ensureOmniInitialized(ctx.cwd);
     ctx.ui.setTitle("Omni-Pi");
     ctx.ui.setTheme(createOmniTheme());
     ctx.ui.setHeader((_tui, theme) => renderHeader(theme));
-    ctx.ui.setStatus("omni", undefined);
+    ctx.ui.setStatus("omni", "\x1b[2mctrl+shift+t tasks\x1b[0m");
   });
 
   api.on("before_agent_start", async (event, ctx) => {

package/extensions/omni-providers/index.ts

@@ -1,5 +1,14 @@
+import type { ExtensionAPI } from "@mariozechner/pi-coding-agent";
+
+import { refreshAuthenticatedProviderModels } from "../../src/model-setup.js";
+import { registerOmniProviders } from "../../src/providers.js";
+
 export default async function omniProvidersExtension(
-  _api: unknown,
+  api: ExtensionAPI,
 ): Promise<void> {
-  return;
+  await registerOmniProviders(api);
+
+  api.on("session_start", async (_event, ctx) => {
+    await refreshAuthenticatedProviderModels(ctx.modelRegistry);
+  });
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "omni-pi",
-  "version": "0.4.0",
+  "version": "0.6.1",
   "description": "Single-agent Pi package that interviews the user, documents the spec, and implements work in bounded slices.",
   "type": "module",
   "license": "MIT",
@@ -30,12 +30,14 @@
   "files": [
     "bin",
     "agents",
+    "CHANGELOG.md",
     "extensions",
     "prompts",
     "skills",
     "src",
     "templates",
     "README.md",
+    "PROVIDERS.md",
     "CREDITS.md"
   ],
   "scripts": {
@@ -44,8 +46,9 @@
     "lint": "biome check .",
     "format": "biome check --write .",
     "test": "vitest run",
+    "verify": "npm run check && npm run lint && npm test && npm run pack:check",
     "pack:check": "npm pack --dry-run",
-    "prepublishOnly": "npm run check && npm run lint && npm test"
+    "prepublishOnly": "npm run verify"
   },
   "devDependencies": {
     "@biomejs/biome": "2.4.9",
@@ -58,7 +61,10 @@
       "./extensions/omni-providers/index.ts",
       "./extensions/omni-core/index.ts",
       "./extensions/omni-memory/index.ts",
-      "./node_modules/pi-web-access/index.ts"
+      "./node_modules/pi-web-access/index.ts",
+      "./node_modules/pi-interview/index.ts",
+      "./node_modules/@juanibiapina/pi-extension-settings/dist/index.js",
+      "./node_modules/@juanibiapina/pi-powerbar/dist"
     ],
     "skills": [
       "./skills",
@@ -70,7 +76,10 @@
   },
   "dependencies": {
     "@anthropic-ai/claude-agent-sdk": "0.2.84",
-    "@mariozechner/pi-coding-agent": "^0.62.0",
+    "@juanibiapina/pi-extension-settings": "^0.6.0",
+    "@juanibiapina/pi-powerbar": "^0.7.1",
+    "@mariozechner/pi-coding-agent": "^0.63.0",
+    "pi-interview": "^0.5.5",
     "pi-subagents": "^0.11.11",
     "pi-web-access": "^0.10.3",
     "zod": "^4.3.6"

package/src/model-command.ts

@@ -1,47 +1,177 @@
+import { mkdir, readFile, writeFile } from "node:fs/promises";
+import path from "node:path";
+
 import type {
   ExtensionAPI,
   ExtensionCommandContext,
 } from "@mariozechner/pi-coding-agent";
+import { getAgentDir } from "@mariozechner/pi-coding-agent";
+
+import { runModelSetupWizard } from "./model-setup.js";
+import { searchableSelect } from "./searchable-select.js";
+
+interface ModelsJsonModel {
+  id: string;
+  [key: string]: unknown;
+}
+
+interface ModelsJsonProvider {
+  models?: ModelsJsonModel[];
+  [key: string]: unknown;
+}
+
+interface ModelsJsonConfig {
+  providers?: Record<string, ModelsJsonProvider>;
+}
 
-import {
-  getAuthenticatedModelOptions,
-  setupCustomProviderModel,
-} from "./model-setup.js";
+type CustomModelEntry = { provider: string; modelId: string };
+type ListOption = { provider: string; modelId: string; label: string };
+
+function getModelsPath(): string {
+  return path.join(getAgentDir(), "models.json");
+}
+
+async function readModelsJson(): Promise<ModelsJsonConfig> {
+  try {
+    const content = await readFile(getModelsPath(), "utf8");
+    const parsed = JSON.parse(content) as ModelsJsonConfig;
+    return typeof parsed === "object" && parsed !== null ? parsed : {};
+  } catch {
+    return {};
+  }
+}
+
+async function writeModelsJson(config: ModelsJsonConfig): Promise<void> {
+  const modelsPath = getModelsPath();
+  await mkdir(path.dirname(modelsPath), { recursive: true });
+  await writeFile(modelsPath, `${JSON.stringify(config, null, 2)}\n`, "utf8");
+}
+
+function getCustomModels(config: ModelsJsonConfig): CustomModelEntry[] {
+  const entries: CustomModelEntry[] = [];
+  for (const [provider, providerConfig] of Object.entries(
+    config.providers ?? {},
+  )) {
+    for (const model of providerConfig.models ?? []) {
+      entries.push({ provider, modelId: model.id });
+    }
+  }
+  return entries;
+}
+
+export function removeCustomModelFromConfig(
+  config: ModelsJsonConfig,
+  provider: string,
+  modelId: string,
+): ModelsJsonConfig {
+  const providers = { ...(config.providers ?? {}) };
+  const providerConfig = providers[provider];
+  if (!providerConfig?.models) {
+    return {
+      ...config,
+      providers,
+    };
+  }
+
+  providerConfig.models = providerConfig.models.filter(
+    (model) => model.id !== modelId,
+  );
+  if (providerConfig.models.length === 0) {
+    delete providers[provider];
+  }
+
+  return {
+    ...config,
+    providers,
+  };
+}
+
+function buildListOptions(customModels: CustomModelEntry[]): ListOption[] {
+  return customModels
+    .map(({ provider, modelId }) => ({
+      provider,
+      modelId,
+      label: `${provider}/${modelId}  ✕`,
+    }))
+    .sort((left, right) => left.label.localeCompare(right.label));
+}
 
 async function handleAdd(ctx: ExtensionCommandContext): Promise<void> {
-  const result = await setupCustomProviderModel({ ctx });
+  const result = await runModelSetupWizard({ ctx });
   ctx.ui.notify(result.summary, "info");
 }
 
 async function handleList(ctx: ExtensionCommandContext): Promise<void> {
-  const models = getAuthenticatedModelOptions(ctx.modelRegistry);
+  const config = await readModelsJson();
+  const custom = getCustomModels(config);
+  const options = buildListOptions(custom);
 
-  if (models.length === 0) {
-    ctx.ui.notify("No models available.", "info");
+  if (options.length === 0) {
+    ctx.ui.notify("No custom models found.", "info");
     return;
   }
 
-  await ctx.ui.select("Available models:", models);
+  while (true) {
+    const selected = await searchableSelect(
+      ctx.ui,
+      "Custom models (✕ = remove):",
+      options.map((option) => ({
+        label: option.label,
+        value: option.label,
+        searchText: option.label.replace("✕", ""),
+      })),
+    );
+    if (selected === undefined) return;
+    const selectedOption = options.find((option) => option.label === selected);
+    if (!selectedOption) continue;
+
+    const modelRef = `${selectedOption.provider}/${selectedOption.modelId}`;
+    const confirmed = await ctx.ui.confirm(
+      "Remove model?",
+      `Remove ${modelRef} from models.json?`,
+    );
+    if (!confirmed) return;
+
+    const freshConfig = await readModelsJson();
+    await writeModelsJson(
+      removeCustomModelFromConfig(
+        freshConfig,
+        selectedOption.provider,
+        selectedOption.modelId,
+      ),
+    );
+
+    ctx.modelRegistry.refresh();
+    ctx.ui.notify(`Removed ${modelRef}.`, "info");
+    return;
+  }
 }
 
 export function registerModelCommand(api: ExtensionAPI): void {
   api.registerCommand("model-setup", {
-    description: "Add or list custom model providers",
+    description: "Add custom providers/models or remove custom model entries",
     async handler(args: string, ctx: ExtensionCommandContext) {
       const sub = args.trim().toLowerCase();
 
       if (sub === "add") return handleAdd(ctx);
       if (sub === "list") return handleList(ctx);
 
-      const choice = await ctx.ui.select("Model setup:", [
-        "add    — Add a custom provider/model",
-        "list   — Show available models",
+      const choice = await searchableSelect(ctx.ui, "Model setup:", [
+        {
+          label: "add    — Add a custom provider or model",
+          value: "add",
+          searchText: "add custom provider model",
+        },
+        {
+          label: "list   — Show custom models / remove model entries",
+          value: "list",
+          searchText: "list remove custom models",
+        },
       ]);
       if (!choice) return;
 
-      const picked = choice.split("—")[0].trim();
-      if (picked === "add") return handleAdd(ctx);
-      if (picked === "list") return handleList(ctx);
+      if (choice === "add") return handleAdd(ctx);
+      if (choice === "list") return handleList(ctx);
     },
   });
 }