test push

.gitignore

@@ -9,7 +9,6 @@ __pycache__/
 
 #temp
 plan-*
-ui
 
 # Distribution / packaging
 .Python

ui/README.md

@@ -0,0 +1,75 @@
+# Agent Workspace UI
+
+A stark, high-contrast monochrome terminal-style frontend for configuring and testing the AI pipeline generation agent.
+
+## Features
+
+- **High-Contrast Monochrome Design**: Stark developer-terminal theme with zero rounded corners, monospaced typography, and red/green accenting reserved strictly for execution outcomes.
+- **In-Memory Session Caching**: Chat history and active model configurations are saved inside the browser's `sessionStorage` (retains logs on page refresh but clears automatically when you close the tab or start a new session).
+- **Collapsible thinking process**: Streamed thinking tokens (`[THINKING PROCESS]`) automatically collapse when finished to keep the logs readable.
+- **Dynamic Tool Execution Cards**: Shows a real-time status spinner, turning green on success (`[OK]`) and collapsing, or red on failure (`[FAILED]`) while showing output details.
+- **Inline Human-In-The-Loop (HITL) Handling**: Allows you to approve/deny bash commands and file writes, or answer clarifying questions directly inside the chat feed.
+- **Local Credentials & Model Catalog**: Saves provider API keys locally, dynamically loads tool-calling models from `models.dev`, and supports adding custom OpenAI-compatible models.
+
+---
+
+## How to Start the System
+
+There are two ways to start the servers: using the single-command helper (recommended) or launching each server manually.
+
+### Option 1: Single-Command Launch (Recommended)
+
+The helper script `ui/run_demo.py` launches both the FastAPI backend and the static UI server in the background, handles clean shutdowns, and verifies backend health before launching your browser.
+
+From the repository root:
+
+```bash
+python ui/run_demo.py
+```
+
+If the browser does not open automatically, or if you prefer to open it yourself, use:
+
+```bash
+python ui/run_demo.py --no-open
+```
+
+You can customize port bindings:
+
+```bash
+python ui/run_demo.py --backend-port 8000 --ui-port 5173
+```
+
+---
+
+### Option 2: Manual Start
+
+If you prefer to run the backend and frontend in separate terminals:
+
+1. **Start the FastAPI Backend**:
+   From the repository root, add `src` to your python path and run `uvicorn`:
+   ```bash
+   # Windows (PowerShell)
+   $env:PYTHONPATH="src"
+   python -m uvicorn agent.api.api:app --host 127.0.0.1 --port 8000
+
+   # Linux/macOS
+   PYTHONPATH=src python -m uvicorn agent.api.api:app --host 127.0.0.1 --port 8000
+   ```
+
+2. **Start the UI Static Server**:
+   From the `ui/` directory, host the static HTML files using python's built-in server:
+   ```bash
+   python -m http.server 5173 --bind 127.0.0.1 --directory ui
+   ```
+
+Once both servers are running, navigate to:
+[http://127.0.0.1:5173/?backend=http://127.0.0.1:8000](http://127.0.0.1:5173/?backend=http://127.0.0.1:8000)
+
+---
+
+## Active API Endpoints consumed by UI
+
+- `POST /generate/stream` — Starts the execution graph for pipeline generation.
+- `POST /generate/{thread_id}/resume/permission` — Resumes a paused execution after human approval or denial of tool execution.
+- `POST /generate/{thread_id}/resume/clarification` — Resumes execution by submitting user answers to clarifying questions.
+- `POST /generate/runs/{run_id}/cancel` — Interrupts and terminates an active pipeline generation run.