Update README.md

README.md

@@ -1,4 +1,4 @@
-YAML
+---
 title: Fastep
 emoji: 🚀
 colorFrom: indigo
@@ -6,72 +6,5 @@ colorTo: purple
 sdk: docker
 app_port: 3100
 pinned: false
----
-
-# Paperclip AI — Local Gemini Pro Edition 🚀
-
-This is a customized, local deployment of the Paperclip AI multi-agent orchestrator. The core connection layer has been modified via a local `opencode` / `codex` binary interceptor to bypass rigid cloud model UI restrictions and seamlessly route all agent intelligence through Google Gemini Pro's free-tier OpenAI compatibility layer.
-
----
-
-## 🛠️ How It Works (Under the Hood)
-
-Paperclip UI enforces strict default model drop-downs (like `openai/gpt-4`) and expects specific machine-readable tool formatting blocks. 
-
-To bypass this without altering the core TypeScript application logic, this setup utilizes a system-level interceptor script:
-1. **Validation Shaking:** When Paperclip initializes or checks for system connectivity, the script instantly returns `hello` to fulfill verification loops.
-2. **Payload Translation:** When a task is run, the script catches the massive system string, extracts the core textual directives, rewrites it into a standard conversational JSON payload, and routes it directly to Google Gemini's endpoint.
-3. **Graceful Fallbacks:** If API rate limits or connection timeouts occur, the script automatically drops back to a beautifully formatted structural markdown roadmap so your local agents never stall or crash the execution loop.
-
----
-
-## 📋 Environment Configuration
-
-To run this environment successfully, ensure your Hugging Face Space or Local Environment contains the following variables configured in your variables and secrets manager:
-
-### Public Variables
-| Variable Name | Value / URL | Purpose |
-| :--- | :--- | :--- |
-| `OPENAI_BASE_URL` | `https://generativelanguage.googleapis.com/v1beta/openai/` | Redirects standard SDK traffic to Google's compatibility engine |
-| `PAPERCLIP_DEFAULT_MODEL`| `gemini-1.5-pro` | Targets the Gemini 1.5 Pro engine profile |
-| `TMPDIR` | `/tmp` | Sets fallback storage directory rules |
-
-### Protected Secrets
-| Secret Name | Value | Purpose |
-| :--- | :--- | :--- |
-| `OPENAI_API_KEY` | `AIzaSy...` | Your unique Google AI Studio Gemini API Key |
-
----
-
-## 👥 Managing the Workforce in the UI
-
-Because the background interceptor overrides incoming configurations globally, follow these exact inputs when building your corporate hierarchy inside the Paperclip dashboard:
-
-1. **Adapter Type:** Set explicitly to `opencode`.
-2. **Model String:** Type or paste exactly `openai/gpt-4` (or any string matching the `provider/model` schema). 
-3. **Command String:** Leave completely blank.
-
-Using this method, you can build out your complete virtual workspace architecture:
-*   **CEO Agent** (Operational coordinator)
-*   **CTO Agent** (Software architect & systems analyst)
-*   **Operations Manager** (Task pipeline manager)
-*   **Product Scout** (Research & market viability)
-
----
-
-## 📊 API Usage Reference (Gemini Free Tier)
-
-Your workspace calls are bound by Google AI Studio's highly sustainable free allowances:
-*   **Requests Per Minute (RPM):** 2 RPM (Perfect for asynchronous multi-agent step iterations)
-*   **Requests Per Day (RPD):** 50 RPD (More than enough for running complete, deep daily code sprints)
-*   **Tokens Per Minute (TPM):** 32,000 TPM
-
----
-
-## 🚀 Launching Your First Sprint
-
-1. Navigate to your Paperclip **Board** tab.
-2. Create or open an active task ticket (e.g., `ATIA-4`).
-3. Move the card column status to **In Progress** or **Doing**.
-4. Post your project brief or directive into the ticket comment section.
-5. Click **Run Task** or allow the background heartbeat system to process the update. Watch the live execution stream output your operational assets directly onto your dashboard panels!
+license: mit
+---