pico-cli 0.1.1__tar.gz → 0.1.2__tar.gz

{pico_cli-0.1.1 → pico_cli-0.1.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: pico-cli
-Version: 0.1.1
+Version: 0.1.2
 Summary: A Python CLI coding agent — autonomous, tool-using, and session-persistent.
 Project-URL: Homepage, https://github.com/Arya-Ojha/PicoCLI_Learn
 Project-URL: Repository, https://github.com/Arya-Ojha/PicoCLI_Learn
@@ -84,6 +84,39 @@ In the TUI, `/provider` opens a picker with ✓/✗ setup status, then a setup f
 
 ## Installation
 
+Requires Python **3.12+**.
+
+```bash
+# recommended: isolated install — `picocli` + `picocli-chat` work in any
+# directory, like `npm i -g`
+pipx install pico-cli
+
+# ... or via uv (same result)
+uv tool install pico-cli
+
+# plain pip also works (see the Windows PATH note below)
+pip install pico-cli
+
+# try without installing (npx-style)
+uvx pico-cli --help
+```
+
+Then open a terminal **in any directory** and run `picocli`. On first launch (no API key yet) the provider setup opens automatically — pick a provider, paste the key, done. Keys can also come from environment variables (see Configuration below).
+
+### Windows: `pip install` and PATH
+
+`pip --user` installs (the default when site-packages isn't writable) drop the commands into `~/.local`-style user Scripts, which is often **not** on PATH — that's when `picocli` is "not recognized". Prefer `pipx`/`uv tool` above, or add the dir once (then open a fresh terminal):
+
+```powershell
+$userPath = [Environment]::GetEnvironmentVariable("Path", "User")
+$newDir = "$env:APPDATA\Python\Python314\Scripts"  # adjust version to yours
+if ($userPath -split ";" -notcontains $newDir) {
+  [Environment]::SetEnvironmentVariable("Path", "$userPath;$newDir", "User")
+}
+```
+
+### From source (contributors)
+
 ```bash
 uv sync
 ```

{pico_cli-0.1.1 → pico_cli-0.1.2}/README.md

@@ -59,6 +59,39 @@ In the TUI, `/provider` opens a picker with ✓/✗ setup status, then a setup f
 
 ## Installation
 
+Requires Python **3.12+**.
+
+```bash
+# recommended: isolated install — `picocli` + `picocli-chat` work in any
+# directory, like `npm i -g`
+pipx install pico-cli
+
+# ... or via uv (same result)
+uv tool install pico-cli
+
+# plain pip also works (see the Windows PATH note below)
+pip install pico-cli
+
+# try without installing (npx-style)
+uvx pico-cli --help
+```
+
+Then open a terminal **in any directory** and run `picocli`. On first launch (no API key yet) the provider setup opens automatically — pick a provider, paste the key, done. Keys can also come from environment variables (see Configuration below).
+
+### Windows: `pip install` and PATH
+
+`pip --user` installs (the default when site-packages isn't writable) drop the commands into `~/.local`-style user Scripts, which is often **not** on PATH — that's when `picocli` is "not recognized". Prefer `pipx`/`uv tool` above, or add the dir once (then open a fresh terminal):
+
+```powershell
+$userPath = [Environment]::GetEnvironmentVariable("Path", "User")
+$newDir = "$env:APPDATA\Python\Python314\Scripts"  # adjust version to yours
+if ($userPath -split ";" -notcontains $newDir) {
+  [Environment]::SetEnvironmentVariable("Path", "$userPath;$newDir", "User")
+}
+```
+
+### From source (contributors)
+
 ```bash
 uv sync
 ```

{pico_cli-0.1.1 → pico_cli-0.1.2}/packages/pico_ai/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pico-cli-ai"
-version = "0.1.1"
+version = "0.1.2"
 description = "LLM abstraction and protocol normalisation."
 readme = "README.md"
 license = "MIT"

{pico_cli-0.1.1 → pico_cli-0.1.2}/packages/pico_core/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pico-cli-core"
-version = "0.1.1"
+version = "0.1.2"
 description = "Agent loop and session tree."
 readme = "README.md"
 license = "MIT"

{pico_cli-0.1.1 → pico_cli-0.1.2}/packages/pico_sdk/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pico-cli-sdk"
-version = "0.1.1"
+version = "0.1.2"
 description = "Headless library API and extension/plugin binding."
 readme = "README.md"
 license = "MIT"

{pico_cli-0.1.1 → pico_cli-0.1.2}/packages/pico_tui/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pico-cli-tui"
-version = "0.1.1"
+version = "0.1.2"
 description = "Terminal user interface view."
 readme = "README.md"
 license = "MIT"

{pico_cli-0.1.1 → pico_cli-0.1.2}/packages/pico_tui/src/pico_tui/app.py

@@ -357,10 +357,20 @@ class PicoApp(App[None]):
         ("f1", "show_help", "Help"),
     ]
 
-    def __init__(self, mgr: _SessionManager, *, startup_notice: str = "") -> None:
+    def __init__(
+        self,
+        mgr: _SessionManager,
+        *,
+        startup_notice: str = "",
+        auto_setup: bool = False,
+    ) -> None:
         super().__init__()
         self._mgr = mgr
         self._startup_notice = startup_notice
+        # First-run onboarding: when True and the active provider is
+        # missing required config, the provider picker opens on launch
+        # so keys/URLs are entered in-UI instead of discovered via /help.
+        self._auto_setup = auto_setup
         self._streaming = False
         # Everything written to the chat log, in order. Thinking blocks are
         # stored as ThinkingSegment so they can collapse/expand on click.
@@ -399,6 +409,10 @@ class PicoApp(App[None]):
             self._write_chat(
                 Panel(self._startup_notice, title="setup", border_style="yellow")
             )
+        if self._auto_setup:
+            settings = self._mgr.session.settings
+            if missing_required(settings.provider, settings):
+                self.call_after_refresh(self._show_provider_picker)
 
     def _refresh_todo_panel(self) -> None:
         """Re-render the todo side panel (auto-hides while empty)."""
@@ -955,7 +969,7 @@ def main(argv: list[str] | None = None) -> int:
             f"Provider {provider_label} is missing required config: "
             f"{', '.join(missing)}. Use /provider to configure it."
         )
-    app = PicoApp(mgr, startup_notice=startup_notice)
+    app = PicoApp(mgr, startup_notice=startup_notice, auto_setup=True)
     app.run()
     return 0
 

{pico_cli-0.1.1 → pico_cli-0.1.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pico-cli"
-version = "0.1.1"
+version = "0.1.2"
 description = "A Python CLI coding agent — autonomous, tool-using, and session-persistent."
 readme = "README.md"
 license = "MIT"

pico_cli-0.1.2/tests/test_first_run_setup.py

@@ -0,0 +1,70 @@
+"""First-run onboarding: missing provider config auto-opens setup."""
+
+import pytest
+
+from pico_sdk.config import Settings
+from pico_sdk.providers import create_provider, missing_required
+from pico_tui.app import PicoApp, _SessionManager
+from pico_tui.provider_picker import ProviderPickerScreen
+
+from conftest import FakeProvider, make_session
+
+
+@pytest.fixture
+def fresh_settings(monkeypatch):
+    """Settings as on a brand-new machine: no keys anywhere."""
+    for var in (
+        "OPENROUTER_API_KEY",
+        "OPENAI_API_KEY",
+        "ANTHROPIC_API_KEY",
+        "GOOGLE_API_KEY",
+        "DEEPSEEK_API_KEY",
+    ):
+        monkeypatch.delenv(var, raising=False)
+    settings = Settings()
+    assert missing_required(settings.provider, settings), "expected missing config"
+    return settings
+
+
+def _app(tmp_path, settings, auto_setup=False):
+    session = make_session(
+        FakeProvider([]), tmp_path, settings=settings, load_skills=False
+    )
+    return PicoApp(_SessionManager(session), auto_setup=auto_setup)
+
+
+async def test_missing_config_auto_opens_provider_picker(tmp_path, fresh_settings):
+    app = _app(tmp_path, fresh_settings, auto_setup=True)
+    async with app.run_test() as pilot:
+        await pilot.pause()
+        await pilot.pause()
+        assert isinstance(app.screen, ProviderPickerScreen)
+
+
+async def test_auto_setup_off_by_default(tmp_path, fresh_settings):
+    app = _app(tmp_path, fresh_settings)
+    async with app.run_test() as pilot:
+        await pilot.pause()
+        assert not isinstance(app.screen, ProviderPickerScreen)
+
+
+async def test_configured_never_auto_opens(tmp_path, monkeypatch):
+    monkeypatch.setenv("OPENROUTER_API_KEY", "test-key")
+    settings = Settings()
+    assert not missing_required(settings.provider, settings)
+    session = make_session(
+        create_provider(settings), tmp_path, settings=settings, load_skills=False
+    )
+    app = PicoApp(_SessionManager(session), auto_setup=True)
+    async with app.run_test() as pilot:
+        await pilot.pause()
+        await pilot.pause()
+        assert not isinstance(app.screen, ProviderPickerScreen)
+
+
+def test_main_requests_auto_setup():
+    import inspect
+
+    from pico_tui.app import main
+
+    assert "auto_setup=True" in inspect.getsource(main)

{pico_cli-0.1.1 → pico_cli-0.1.2}/uv.lock

@@ -408,7 +408,7 @@ wheels = [
 
 [[package]]
 name = "pico-cli"
-version = "0.1.1"
+version = "0.1.2"
 source = { editable = "." }
 dependencies = [
     { name = "pico-cli-sdk" },
@@ -437,7 +437,7 @@ dev = [
 
 [[package]]
 name = "pico-cli-ai"
-version = "0.1.1"
+version = "0.1.2"
 source = { editable = "packages/pico_ai" }
 dependencies = [
     { name = "httpx" },
@@ -452,7 +452,7 @@ requires-dist = [
 
 [[package]]
 name = "pico-cli-core"
-version = "0.1.1"
+version = "0.1.2"
 source = { editable = "packages/pico_core" }
 dependencies = [
     { name = "httpx" },
@@ -469,7 +469,7 @@ requires-dist = [
 
 [[package]]
 name = "pico-cli-sdk"
-version = "0.1.1"
+version = "0.1.2"
 source = { editable = "packages/pico_sdk" }
 dependencies = [
     { name = "pico-cli-ai" },
@@ -486,7 +486,7 @@ requires-dist = [
 
 [[package]]
 name = "pico-cli-tui"
-version = "0.1.1"
+version = "0.1.2"
 source = { editable = "packages/pico_tui" }
 dependencies = [
     { name = "pico-cli-sdk" },

pico_cli-0.1.1/IMPLEMENTATION_SUMMARY.md

@@ -1,80 +0,0 @@
-# Context Status Bar Implementation Summary
-
-## Overview
-Added a context status bar widget to the PicoCLI TUI that displays:
-- Provider name (e.g., "OpenRouter")
-- Model name (e.g., "nvidia/nemotron-3.5-lightning:free")
-- Thinking indicator (shown when the model is reasoning)
-- Context window progress bar (fills as context fills, color-coded by usage level)
-- Current token count (formatted with commas, e.g., "38,932")
-
-## Files Modified
-
-### 1. `packages/pico_core/src/pico_core/fsm.py`
-- Added public `estimate_tokens()` method to `AgentLoop` class
-- This exposes the private `_estimate_tokens()` functionality
-
-### 2. `packages/pico_sdk/src/pico_sdk/session.py`
-- Added `provider_name` property to `AgentSession` class
-  - Extracts provider name from the provider class name
-  - Returns "OpenRouter" for OpenRouterProvider
-- Added `context_window` property to `AgentSession` class
-  - Returns the context window size from the agent loop
-- Added `estimate_tokens()` method to `AgentSession` class
-  - Returns current estimated token count from the agent loop
-
-### 3. `packages/pico_tui/src/pico_tui/status_bar.py` (NEW FILE)
-- Created `ContextStatusBar` widget extending Textual's `Static`
-- Features:
-  - Displays provider | model information
-  - Shows "thinking" indicator in yellow italic when active
-  - Renders a 20-character progress bar for context window usage
-  - Color-codes the progress bar:
-    - Green: 0-50% usage (░ character)
-    - Yellow: 50-70% usage (▒ character)
-    - Orange: 70-90% usage (▓ character)
-    - Red: 90-100% usage (█ character)
-  - Shows token count with comma formatting in cyan
-- Methods:
-  - `on_mount()`: Renders initial display after widget mount
-  - `update_info()`: Updates all status information
-  - `set_thinking()`: Updates thinking state
-  - `_update_display()`: Renders the status bar with Rich Text
-
-### 4. `packages/pico_tui/src/pico_tui/app.py`
-- Imported `ContextStatusBar` widget
-- Added CSS styling for `#status-bar`:
-  - Docked to bottom
-  - Height: 1 line
-  - Background: $surface color
-  - Padding: 0 1
-- Added `ContextStatusBar` to `compose()` method after Input widget
-- Added `_update_status_bar()` method to update the status bar with current session info
-- Updated `on_mount()` to call `_update_status_bar()` on initialization
-- Updated `_run_prompt()` to call `_update_status_bar()` when streaming starts
-- Updated `_stream_worker()` to call `_update_status_bar()` when streaming completes
-
-### 5. `tests/test_status_bar.py` (NEW FILE)
-- Added 3 tests for the ContextStatusBar widget:
-  - `test_status_bar_initialization`: Verifies default values
-  - `test_status_bar_stores_info`: Verifies info storage
-  - `test_status_bar_set_thinking`: Verifies thinking state updates
-
-## Test Results
-All 92 tests pass successfully, including:
-- 19 existing TUI tests
-- 3 new status bar tests
-- All other existing tests remain passing
-
-## Usage
-The status bar automatically appears below the input bar in the TUI and updates:
-- On application startup (shows initial provider, model, and 0 tokens)
-- When a prompt is submitted (shows "thinking" indicator)
-- When streaming completes (updates token count and hides "thinking")
-- When the model changes via `/model` command
-
-The status bar provides real-time visibility into:
-- Which AI provider and model is being used
-- Whether the model is currently thinking/reasoning
-- How much of the context window is being used
-- The exact token count in the current context