🤖 NashAgents

In-house CLI agentic tool for NashTech — define, manage, and invoke AI agents from anywhere.

NashAgents is a pip-installable CLI tool that lets your team run pre-built domain-specific AI agents or create custom ones through simple YAML config files.

---

⚡ Quick Start

# Install
cd cli_agentic_tool_tryout
pip install -e .

# See all agents
nashagents list

# Run an agent
nashagents run backend-dev

# Create a custom agent
nashagents create my-agent --interactive

📦 Installation

From source (development)

git clone <repo-url>
cd cli_agentic_tool_tryout
pip install -e .

As a package (production)

pip install nashagents

🔑 Configuration

1. Set up your LLM provider

Copy the env template and add your API key:

cp .env.example .env
# Edit .env with your keys

Supported providers (via litellm):

Provider	Environment Variable	Model Format
OpenAI	OPENAI_API_KEY	openai/gpt-4o
Azure OpenAI	AZURE_API_KEY + AZURE_API_BASE	azure/gpt-4o
Google Gemini	GEMINI_API_KEY	gemini/gemini-pro
Anthropic	ANTHROPIC_API_KEY	anthropic/claude-3-sonnet

2. Initialize config

nashagents init

This creates ~/.nashagents/ with default configuration.

3. Set a default model

nashagents config --set-model openai/gpt-4o

---

🤖 Pre-built Agents

Agent	Description	Tags
backend-dev	Backend APIs, databases, microservices	development, backend, python
frontend-dev	React, Angular, Vue, CSS, responsive UI	development, frontend, ui
devops	Docker, K8s, CI/CD, cloud, IaC	infrastructure, devops, cloud
qa-engineer	Test strategies, automation, quality	testing, qa, automation
technical-writer	Docs, APIs, READMEs, tutorials	writing, documentation
code-reviewer	Code quality, security, performance	review, security, quality
data-engineer	ETL, SQL, pipelines, analytics	data, etl, sql, pipeline
architect	System design, HLD/LLD, tech stack	architecture, design

---

📋 CLI Commands

List agents

nashagents list                  # All agents
nashagents list --tag backend    # Filter by tag

Run an agent

nashagents run backend-dev                          # Interactive chat
nashagents run backend-dev --model openai/gpt-4o    # Override model
nashagents run code-reviewer --message "Review app.py"  # Single shot
nashagents run frontend-dev --no-stream             # Disable streaming

Interactive session commands

Command	Action
/help	Show available commands
/model	Show current model
/tokens	Show message count
/clear	Clear conversation history
/agent	Show agent details
/quit	Exit session

Create a custom agent

# Quick create
nashagents create my-agent -d "My custom agent" -s "You are a helpful assistant"

# Interactive wizard
nashagents create my-agent --interactive

# With all options
nashagents create security-auditor \
  --description "Security audit specialist" \
  --system-prompt "You are a cybersecurity expert..." \
  --model openai/gpt-4o \
  --tools shell,file_read \
  --tags security,audit

Manage agents

nashagents info backend-dev      # Show agent details
nashagents delete my-agent       # Delete custom agent
nashagents config --show         # Show config info

---

📝 Custom Agent Config (YAML)

Create custom agents by adding YAML files to ~/.nashagents/agents/:

name: "security-auditor"
description: "Security audit specialist for code and infrastructure"
version: "1.0"

system_prompt: |
  You are a cybersecurity expert specializing in application security.
  Analyze code for vulnerabilities, misconfigurations, and security risks.
  Follow OWASP guidelines and provide remediation steps.

model: "openai/gpt-4o"      # Optional: override default model
temperature: 0.2             # Low for precise analysis
max_tokens: 4096

tools:                       # Available: shell, file_read, file_write, web_search
  - shell
  - file_read

tags:
  - security
  - audit
  - compliance

---

🔧 Available Tools

Tool	Description
shell	Execute shell commands (tests, builds, git, etc.)
file_read	Read file contents
file_write	Write/create files
web_search	Search the web (requires API key)

---

🏗️ Project Structure

nashagents/
├── __init__.py          # Package metadata
├── cli.py               # CLI commands (Click)
├── config.py            # YAML config management
├── runner.py            # Agent execution engine
├── tools.py             # Built-in tool implementations
├── ui.py                # Rich terminal UI
└── builtin_agents/      # Pre-built agent YAML configs
    ├── backend_dev.yaml
    ├── frontend_dev.yaml
    ├── devops.yaml
    ├── qa_engineer.yaml
    ├── technical_writer.yaml
    ├── code_reviewer.yaml
    ├── data_engineer.yaml
    └── architect.yaml

---

🤝 Adding Agents for Your Team

1. Create a YAML config in ~/.nashagents/agents/
2. Share configs via your org's git repo or internal package registry
3. Override built-in agents by creating a custom agent with the same name

---

📄 License

MIT