RustAIgent

RustAIgent is a command‑line coding assistant written in Rust. It leverages large language models (LLMs) to provide interactive code assistance, file and system operations, and more. By exposing rich function‑calling tools, RustAIgent enables the LLM to:

• Inspect and modify local files
• Run shell commands
• Fetch remote resources
• Compile and evaluate code snippets
• Handle batch prompts concurrently

It supports multiple AI backends (OpenAI, Anthropic/Claude, Ollama, Google Generative API), with robust retry and batching logic.

---

Table of Contents

1. Architecture
2. Features
3. Installation
4. Configuration
5. Usage
6. Examples
7. Advanced Usage
8. Environment Variables
9. Contribution
10. Roadmap
11. License

---

Architecture

RustAIgent follows a modular design:

1. Core Agent
   • Manages conversation state, tool definitions, and dispatch logic.
   • Routes requests to the configured provider (OpenAI, Claude, Ollama, or Google).
2. Function Calling Layer
   • Defines a set of JSON‑schema–based tools (read_file, write_file, delete_file, list_dir, run_command, fetch_url, eval_code).
   • Automatically detects and executes tool calls from LLM responses.
3. Provider Integrations
   • OpenAI: Chat Completions API with function calling.
   • Anthropic (Claude): Text completion via the Anthropic API.
   • Ollama: Local LLM endpoint (e.g. localhost:11434).
   • Google: Generative Language API (Chat Bison).
4. Reliability & Scalability
   • Retry Mechanism: Configurable exponential backoff for failed API calls.
   • Batching: Parallel prompt processing using Tokio tasks and send_batch_requests.

---

Features

• File I/O: read_file(path), write_file(path, content), delete_file(path)
• Filesystem Operations: list_dir(path)
• Shell Execution: run_command(command)
• HTTP Fetching: fetch_url(url)
• Code Evaluation: eval_code(code) (stub for secure compilation)
• Multi‑Provider: Choose between openai, claude, ollama, google
• Retries & Backoff: Controlled via RETRY_COUNT and BACKOFF_BASE_MS
• Batch Requests: Process multiple prompts concurrently
• Customizable: MODEL_NAME, MAX_TOKENS, TEMPERATURE via env vars

---

Installation

1. Ensure you have Rust (1.60+) installed. If not, install via rustup.
2. Clone the repository:

   git clone https://github.com/makalin/RustAIgent.git
   cd RustAIgent

3. Build in release mode:

   cargo build --release

4. The binary will be at ./target/release/RustAIgent.

---

Configuration

RustAIgent reads configuration from environment variables (or a .env file). See Environment Variables.

---

Usage

Run the agent and interact via standard input:

./target/release/RustAIgent

Switch providers on the fly:

API_PROVIDER=claude ./target/release/RustAIgent
API_PROVIDER=google ./target/release/RustAIgent

During the session, prefix commands to invoke tools explicitly, or let the model choose automatically:

You: read_file("config.toml")
You: run_command("cargo fmt -- --check")
You: fetch_url("https://example.com/data.json")

---

Examples

1. Reading a file

   You: read_file("src/main.rs")
   RustAIgent: [TOOL] -- Contents of src/main.rs...
   

2. Listing directory

   You: list_dir(".")
   RustAIgent: [TOOL] -- src, Cargo.toml, README.md, ...
   

3. Writing to a file

   You: write_file("note.txt", "This is a test.")
   RustAIgent: File written successfully.
   

4. Running a shell command

   You: run_command("ls -la target/release")
   RustAIgent: target binary, debug symbols, etc.
   

---

Advanced Usage

Batch Processing

Use send_batch_requests to handle multiple prompts concurrently in code:

let prompts = vec!["List files in /etc".into(), "Fetch Rust docs".into()];
let responses = agent.send_batch_requests(prompts).await?;
for msg in responses { println!("Response: {}", msg.content); }

Custom Retry Strategy

Adjust retry parameters in .env:

RETRY_COUNT=5
BACKOFF_BASE_MS=200

Custom Model & Temperature

MODEL_NAME=gpt-4o-mini
TEMPERATURE=0.3
MAX_TOKENS=512

---

Environment Variables

Variable	Description	Default
OPENAI_API_KEY	API key for OpenAI	required
GOOGLE_API_KEY	API key for Google Generative API	optional
API_PROVIDER	openai, claude, ollama, or google	openai
MODEL_NAME	Model identifier for provider	gpt-4o-mini
MAX_TOKENS	Maximum tokens per completion	1024
TEMPERATURE	Sampling temperature (0.0–1.0)	0.7
RETRY_COUNT	Number of retry attempts on failure	3
BACKOFF_BASE_MS	Base backoff duration in ms	500

---

Contribution

1. Fork this repository
2. Create a new branch: git checkout -b feature/awesome
3. Commit your changes: git commit -am 'Add awesome feature'
4. Push to the branch: git push origin feature/awesome
5. Create a Pull Request

Please follow the Rust style guidelines and include tests where appropriate.

---

Roadmap

• Secure sandbox for eval_code
• Enhanced function parameter validation
• Jittered exponential backoff
• Support for additional LLM providers
• Plugin architecture for custom tools

---

License

This project is licensed under the MIT License. See the LICENSE file for details.