Skip to content

A curated collection of practical AI projects implementing OCR systems, RAG, AI agents, and other AI use cases.

chore: Resolve merge conflict and organize Eagle Eye into subfolder

Tiioluwanid5ebe0a72d5a
diff --cc README.md
index 3ae8123,bfe45af..b9dbe5e
--- a/README.md
+++ b/README.md
@@@ -1,100 -1,135 +1,17 @@@
- # πŸ¦… Eagle Eye
-
-# Eagle Eye
-
-AI-powered GitHub PR review agent using OpenClaw, MiniMax M2.7, and GitHub MCP. Triggered via Telegram.
-
-## Project Overview
-
-Eagle Eye is a Telegram-triggered code review assistant that analyses GitHub pull requests and delivers structured feedback directly to your Telegram chat. It uses MiniMax M2.7 as the reasoning model and the GitHub MCP server to fetch PR diffs and post reviews. Reviews cover security vulnerabilities, bugs, code quality, and best practices, with findings grouped by severity and an overall rating. Nothing is posted to GitHub without your explicit approval.
-
----
-
-## How It Works
-
-![Eagle Eye Architecture](docs/architecture.svg)
-
----
-
-## Prerequisites
-
-- **[OpenClaw](https://openclaw.dev)** installed and available in your PATH
-- **Telegram bot token**: create a bot via [@BotFather](https://t.me/BotFather) and note the token
-- **GitHub Personal Access Token (PAT)**: with `repo`, `pull_requests: write`, and `issues: write` permissions
-
----
-
-## Setup
-
-### 1. Run `openclaw onboard`
-
-```bash
-openclaw onboard
-```
-
-When prompted, select **MiniMax M2.7** as your model. OpenClaw will ask for your API key and store it securely in its auth profiles; no need to put it in `openclaw.json`.
-
-### 2. Configure `openclaw.json`
-
-Copy `openclaw.example.json` to `openclaw.json` and fill in your credentials:
-
-```json
-{
- "gateway": {
- "mode": "local",
- "auth": {
- "mode": "token",
- "token": "YOUR_OPENCLAW_GATEWAY_TOKEN"
- }
- },
- "channels": {
- "telegram": {
- "enabled": true,
- "botToken": "YOUR_TELEGRAM_BOT_TOKEN"
- }
- },
- "mcp": {
- "servers": {
- "github": {
- "command": "npx",
- "args": ["-y", "@modelcontextprotocol/server-github"],
- "env": {
- "GITHUB_PERSONAL_ACCESS_TOKEN": "YOUR_GITHUB_PAT"
- }
- }
- }
- }
-}
-```
-
-### 3. Copy the project to your OpenClaw workspace
-
-**macOS / Linux:**
-```bash
-cp -r . ~/.openclaw/workspace/skills/eagle-eye/
-```
-
-**Windows:**
-```cmd
-xcopy . %USERPROFILE%\.openclaw\workspace\skills\eagle-eye\ /E /I
-```
-
-### 4. Start the OpenClaw gateway
-
-```bash
-openclaw gateway
-```
-
-The agent is now listening for messages on Telegram.
-
----
-
-## Usage
-
-Send any message containing a GitHub PR URL to your Telegram bot:
-
-```
-https://github.com/owner/repo/pull/42
-```
-
-Or inline with other text:
-
-```
-Please review this: https://github.com/owner/repo/pull/42
-```
-
-The agent fetches the PR, analyses the diff, and sends a structured review back to Telegram. It then asks whether you want to post the review to GitHub: reply **`post`** to post it as a GitHub comment, or **`no`** to discard it.
-
-You can also reply with edits or instructions (e.g. *"add a note about the missing tests"*) and the agent will revise the review and ask again.
-
----
-
-## Known Limitations
-
-- **No CI/CD context**: the agent cannot see test results, build logs, or coverage reports unless you paste them into the chat
-- **Large PRs**: for PRs with 200+ changed files, the review prioritises security and correctness findings and may not cover every file exhaustively
-- **No session memory**: the agent does not remember previous reviews between Telegram sessions; each conversation starts fresh
-- **Single reviewer identity**: the agent posts as whichever GitHub account owns the PAT; it cannot impersonate other reviewers
-=======
+ # **Hands-On AI Engineering** πŸš€
- AI-powered GitHub PR review agent using [OpenClaw](https://openclaw.dev), MiniMax M2.7, and GitHub MCP. Triggered via Telegram.
+ A curated repository of AI-powered applications and agentic systems showcasing practical use cases of Large Language Models (LLMs) from providers like Google, Anthropic, OpenAI, and self-hosted open-source models.
- ---
+ ## **What’s Inside?**
- ## πŸ“½οΈ Project Overview
+ * πŸ€– **AI Agents & Use Cases** – Explore a variety of agent-based AI applications.
+ * πŸ“š **RAG (Retrieval-Augmented Generation)** – Implementations of knowledge-enhanced AI models.
+ * πŸš€ **Scalable AI Solutions** – Best practices for building production-ready AI applications.
- **Eagle Eye** is a Telegram-triggered code review assistant that analyzes GitHub pull requests and delivers structured feedback directly to your chat.
+ ## **Contributing**
- By leveraging the **MiniMax M2.7** reasoning model and the **GitHub MCP** server, it fetches PR diffs, identifies security vulnerabilities or bugs, and provides an overall quality rating. Critically, the agent acts as your co-pilot: it prepares the review, but **nothing is posted to GitHub without your explicit approval.**
+ We welcome contributions! If you have an AI app, agent, or enhancement to share, check the **Issues** section and submit a pull request.
- ---
+ ## **πŸ“œ License**
- ## πŸ—οΈ Project Structure
-
- Understanding the layout of the Eagle Eye agent:
-
- ```text
- .
- β”œβ”€β”€ .claude/ # Claude-specific IDE settings
- β”œβ”€β”€ docs/ # Documentation assets
- β”‚ └── architecture.svg # System architecture diagram
- β”œβ”€β”€ openclaw.json # Your active configuration (Telegram, MCP)
- β”œβ”€β”€ openclaw.example.json # Template for local configuration
- β”œβ”€β”€ README.md # You are here
- └── SOUL.md # 🧠 The agent's core identity and review logic
- ```
-
- ### 🧠 The SOUL (`SOUL.md`)
- The `SOUL.md` file is the most important part of the agent. While traditional software hardcodes logic in scripts, Eagle Eye uses this "Soul" file to define its:
- - **Identity & Tone**: How it speaks (professional, constructive, concise).
- - **Review Criteria**: Specific focus on Security (XSS, Injection), Bugs (Race conditions, logic errors), and Best Practices.
- - **Workflow**: The step-by-step instructions for fetching data, analyzing diffs, and formatting the Telegram response.
-
- ---
-
- ## βš™οΈ How It Works
-
- ![Eagle Eye Architecture](docs/architecture.svg)
-
- 1. **Trigger**: You send a GitHub PR URL to your Telegram bot.
- 2. **Fetch**: The agent uses the **GitHub MCP Server** to securely pull the PR description and code diff.
- 3. **Analysis**: The **MiniMax M2.7** model processes the diff based on the instructions in `SOUL.md`.
- 4. **Draft**: A formatted review is sent to your Telegram chat with a summary and severity-coded findings.
- 5. **Approval**: You reply with `post` to publish it as a comment on GitHub, or `no` to discard it.
-
- ---
-
- ## πŸš€ Setup
-
- ### 1. Run `openclaw onboard`
-
- ```bash
- openclaw onboard
- ```
-
- When prompted, select **MiniMax M2.7** as your model. OpenClaw will ask for your API key and store it securely in its auth profiles; **no need to put it in `openclaw.json`**.
-
- ### 2. Configure `openclaw.json`
- Copy `openclaw.example.json` to `openclaw.json` and fill in your credentials (Telegram Bot Token and GitHub PAT).
-
- ### 3. Deploy to Workspace
- Copy the project files into your OpenClaw skills directory:
-
- **Windows:**
- ```powershell
- xcopy . %USERPROFILE%\.openclaw\workspace\skills\eagle-eye\ /E /I
- ```
-
- **macOS / Linux:**
- ```bash
- cp -r . ~/.openclaw/workspace/skills/eagle-eye/
- ```
-
- ### 4. Start the Agent
- Run the gateway to begin listening for messages:
- ```bash
- openclaw gateway
- ```
-
- ---
-
- ## πŸ“± Usage
-
- Simply paste a PR link into your Telegram chat:
- > *Please review this: https://github.com/owner/repo/pull/42*
-
- The agent will respond with the review and a prompt:
- - Reply **`post`** to finalize the review on GitHub.
- - Reply **`no`** to cancel.
- - Reply with **instructions** (e.g., *"Make it more concise"*) to iterate on the draft.
-
- ---
-
- ## ⚠️ Known Limitations
-
- - **Scope**: Does not see CI/CD logs or test results unless pasted into the chat.
- - **Large PRs**: For very large changes, the agent prioritizes security and correctness.
- - **Stateless**: Each Telegram session starts fresh; previous review context is not retained unless manually provided.
+ This repository is licensed under the **MIT License**. See the [LICENSE](./LICENSE) file for details.
-
diff --cc eagle-eye/README.md
index 0000000,55c6b1c..3ae8123
mode 000000,100644..100644
--- a/eagle-eye/README.md
+++ b/eagle-eye/README.md
@@@ -1,0 -1,115 +1,100 @@@
-# Eagle Eye
++# πŸ¦… Eagle Eye
+
-AI-powered GitHub PR review agent using OpenClaw, MiniMax M2.7, and GitHub MCP. Triggered via Telegram.
++AI-powered GitHub PR review agent using [OpenClaw](https://openclaw.dev), MiniMax M2.7, and GitHub MCP. Triggered via Telegram.
+
-## Project Overview
++---
++
++## πŸ“½οΈ Project Overview
+
-Eagle Eye is a Telegram-triggered code review assistant that analyses GitHub pull requests and delivers structured feedback directly to your Telegram chat. It uses MiniMax M2.7 as the reasoning model and the GitHub MCP server to fetch PR diffs and post reviews. Reviews cover security vulnerabilities, bugs, code quality, and best practices, with findings grouped by severity and an overall rating. Nothing is posted to GitHub without your explicit approval.
++**Eagle Eye** is a Telegram-triggered code review assistant that analyzes GitHub pull requests and delivers structured feedback directly to your chat.
++
++By leveraging the **MiniMax M2.7** reasoning model and the **GitHub MCP** server, it fetches PR diffs, identifies security vulnerabilities or bugs, and provides an overall quality rating. Critically, the agent acts as your co-pilot: it prepares the review, but **nothing is posted to GitHub without your explicit approval.**
+
+ ---
+
-## How It Works
++## πŸ—οΈ Project Structure
+
-![Eagle Eye Architecture](docs/architecture.svg)
++Understanding the layout of the Eagle Eye agent:
++
++```text
++.
++β”œβ”€β”€ .claude/ # Claude-specific IDE settings
++β”œβ”€β”€ docs/ # Documentation assets
++β”‚ └── architecture.svg # System architecture diagram
++β”œβ”€β”€ openclaw.json # Your active configuration (Telegram, MCP)
++β”œβ”€β”€ openclaw.example.json # Template for local configuration
++β”œβ”€β”€ README.md # You are here
++└── SOUL.md # 🧠 The agent's core identity and review logic
++```
++
++### 🧠 The SOUL (`SOUL.md`)
++The `SOUL.md` file is the most important part of the agent. While traditional software hardcodes logic in scripts, Eagle Eye uses this "Soul" file to define its:
++- **Identity & Tone**: How it speaks (professional, constructive, concise).
++- **Review Criteria**: Specific focus on Security (XSS, Injection), Bugs (Race conditions, logic errors), and Best Practices.
++- **Workflow**: The step-by-step instructions for fetching data, analyzing diffs, and formatting the Telegram response.
+
+ ---
+
-## Prerequisites
++## βš™οΈ How It Works
++
++![Eagle Eye Architecture](docs/architecture.svg)
+
-- **[OpenClaw](https://openclaw.dev)** installed and available in your PATH
-- **Telegram bot token**: create a bot via [@BotFather](https://t.me/BotFather) and note the token
-- **GitHub Personal Access Token (PAT)**: with `repo`, `pull_requests: write`, and `issues: write` permissions
++1. **Trigger**: You send a GitHub PR URL to your Telegram bot.
++2. **Fetch**: The agent uses the **GitHub MCP Server** to securely pull the PR description and code diff.
++3. **Analysis**: The **MiniMax M2.7** model processes the diff based on the instructions in `SOUL.md`.
++4. **Draft**: A formatted review is sent to your Telegram chat with a summary and severity-coded findings.
++5. **Approval**: You reply with `post` to publish it as a comment on GitHub, or `no` to discard it.
+
+ ---
+
-## Setup
++## πŸš€ Setup
+
+ ### 1. Run `openclaw onboard`
+
+ ```bash
+ openclaw onboard
+ ```
+
-When prompted, select **MiniMax M2.7** as your model. OpenClaw will ask for your API key and store it securely in its auth profiles; no need to put it in `openclaw.json`.
++When prompted, select **MiniMax M2.7** as your model. OpenClaw will ask for your API key and store it securely in its auth profiles; **no need to put it in `openclaw.json`**.
+
+ ### 2. Configure `openclaw.json`
++Copy `openclaw.example.json` to `openclaw.json` and fill in your credentials (Telegram Bot Token and GitHub PAT).
+
-Copy `openclaw.example.json` to `openclaw.json` and fill in your credentials:
-
-```json
-{
- "gateway": {
- "mode": "local",
- "auth": {
- "mode": "token",
- "token": "YOUR_OPENCLAW_GATEWAY_TOKEN"
- }
- },
- "channels": {
- "telegram": {
- "enabled": true,
- "botToken": "YOUR_TELEGRAM_BOT_TOKEN"
- }
- },
- "mcp": {
- "servers": {
- "github": {
- "command": "npx",
- "args": ["-y", "@modelcontextprotocol/server-github"],
- "env": {
- "GITHUB_PERSONAL_ACCESS_TOKEN": "YOUR_GITHUB_PAT"
- }
- }
- }
- }
-}
-```
++### 3. Deploy to Workspace
++Copy the project files into your OpenClaw skills directory:
+
-### 3. Copy the project to your OpenClaw workspace
++**Windows:**
++```powershell
++xcopy . %USERPROFILE%\.openclaw\workspace\skills\eagle-eye\ /E /I
++```
+
+ **macOS / Linux:**
+ ```bash
+ cp -r . ~/.openclaw/workspace/skills/eagle-eye/
+ ```
+
-**Windows:**
-```cmd
-xcopy . %USERPROFILE%\.openclaw\workspace\skills\eagle-eye\ /E /I
-```
-
-### 4. Start the OpenClaw gateway
-
++### 4. Start the Agent
++Run the gateway to begin listening for messages:
+ ```bash
+ openclaw gateway
+ ```
+
-The agent is now listening for messages on Telegram.
-
+ ---
+
-## Usage
-
-Send any message containing a GitHub PR URL to your Telegram bot:
-
-```text
-https://github.com/owner/repo/pull/42
-```
-
-Or inline with other text:
-
-```text
-Please review this: https://github.com/owner/repo/pull/42
-```
++## πŸ“± Usage
+
-The agent fetches the PR, analyses the diff, and sends a structured review back to Telegram. It then asks whether you want to post the review to GitHub: reply **`post`** to post it as a GitHub comment, or **`no`** to discard it.
++Simply paste a PR link into your Telegram chat:
++> *Please review this: https://github.com/owner/repo/pull/42*
+
-You can also reply with edits or instructions (e.g. *"add a note about the missing tests"*) and the agent will revise the review and ask again.
++The agent will respond with the review and a prompt:
++- Reply **`post`** to finalize the review on GitHub.
++- Reply **`no`** to cancel.
++- Reply with **instructions** (e.g., *"Make it more concise"*) to iterate on the draft.
+
+ ---
+
-## Known Limitations
++## ⚠️ Known Limitations
+
-- **No CI/CD context**: the agent cannot see test results, build logs, or coverage reports unless you paste them into the chat
-- **Large PRs**: for PRs with 200+ changed files, the review prioritises security and correctness findings and may not cover every file exhaustively
-- **No session memory**: the agent does not remember previous reviews between Telegram sessions; each conversation starts fresh
-- **Single reviewer identity**: the agent posts as whichever GitHub account owns the PAT; it cannot impersonate other reviewers
++- **Scope**: Does not see CI/CD logs or test results unless pasted into the chat.
++- **Large PRs**: For very large changes, the agent prioritizes security and correctness.
++- **Stateless**: Each Telegram session starts fresh; previous review context is not retained unless manually provided.