No project description provided
Project description
Ainator
Ainator is a CLI/API framework for AI agents built on the agno framework. It provides interactive chat, persistent sessions, knowledge bases (RAG), and built-in tools for shell execution, file operations, and web search.
Features
- Interactive Chat: Persistent sessions with compression and summarization
- Knowledge Bases (RAG): Embed code repositories (AST-aware chunking), websites, and directories
- Built-in Tools: Shell execution (with safety controls), file operations, DuckDuckGo search
- Session Management: List, switch, show, and compress sessions
- Learning & Memory: User profile, memory, and session context
- Cultural Knowledge: Context-aware cultural knowledge integration
- KeyboardInterrupt Handling: Proper handling of interruption signals
- Skills System: Extensible via
.ainator/skills/directory - Web Server: FastAPI server with Swagger documentation
- Model Support: VLLM, DeepSeek, OpenAI, Anthropic...
Installation
pip install ainator
Quick Start
# Run a one-off prompt (creates new session)
ainator hi "What is the capital of France?"
# Continue in existing session
ainator but "Tell me more about that"
# Or start interactive chat
ainator chat
# Or Start the web server (localhost:7777/docs)
ainator server
# Then consume it through https://github.com/agno-agi/agent-ui
Commands
Chat & Prompt
ainator hi "your prompt"- One-off prompt execution (creates new session)ainator but "your prompt"- Continue in existing sessionainator chat- Interactive chat session
Test-Driven Development & Fixing
ainator fix "python my_script.py or any command or pytest"- Run until rc == 0
Session Management
ainator sessions- List all sessionsainator switch <id>- Switch to a sessionainator show- Show messages in current session
Skills Management
ainator skills- List available skills
RAG Management
ainator rag code add ./path/to/repo python- Add code repositoryainator rag website add <url>- Add website contentainator rag search <knowledge-name> "query"- Search knowledge baseainator rag remove <knowledge-name>- Remove knowledge baseainator rag- Show RAG status
Server
ainator server- Start FastAPI web server (port 7777)
Configuration
Configuration file is auto-managed, but you can still hack it if you want to
tweak parameters: .ainator/config.yml
Example Configuration
model: agno.models.vllm:VLLM id=coder
db: agno.db.sqlite:SqliteDb
tools:
- FileTools
- ShellTools
- DuckDuckGoTools
knowledge:
- name: ainator
path: /home/jpic/src/ainator
plugin: code
agent:
reasoning: true
add_culture_to_context: true
store_events: true
learning:
user_profile: true
user_memory: true
session_context: true
Model Configuration Examples
# VLLM
model: agno.models.vllm:VLLM id=coder
# DeepSeek
model: agno.models.deepseek:DeepSeek id=deepseek-chat
Environment Variables
AINATOR_MODEL- Override the model configuration
Architecture
Core Framework
- agno: Core AI agent framework (agents, models, tools, SQLite database)
- cli2: Command-line interface framework
Knowledge Plugins
- code: Repository embedding with AST-aware chunking
- Uses llama-index CodeSplitter and HierarchicalNodeParser
- BAAI/bge-small-en-v1.5 embeddings
- LanceDB vector storage
- site: Website scraping with Parsel
- generic: Directory embedding descriptive file names is required
Skills System
Extensible via .ainator/skills/ directory, just create a skill in there or
clone a skills repo such as:
git clone https://github.com/anthropics/skills .ainator/skills/anthropics
Renderer
Streaming Markdown output
Server
FastAPI server via AgentOS, consumable with AgentUI or the AgentOS client, or plain HTTP even.
Tools
- Enhanced shell execution with chunked output
- File operations
- DuckDuckGo search
- Custom knowledge tools (RAGs)
Compression
Automatic
Learning
- User profile management
- Memory system
- Session context tracking
Development
Installation for Development
pip install -e '.'
Testing
pytest tests/
Version
Version 0.0.3 (defined in src/ainator/__about__.py)
License
MIT © 2026-present jpic
Links
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
File details
Details for the file ainator-0.0.6.tar.gz.
File metadata
- Download URL: ainator-0.0.6.tar.gz
- Upload date:
- Size: 17.5 MB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.14.2
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
64229ebcfa10c97b9f16bcad85e902cb113d6ef33e55c375fbfd33638b6f19db
|
|
| MD5 |
b9b90a90882c279cdd868da7ad47e3ec
|
|
| BLAKE2b-256 |
d723cbd04feb225fa44f826bd93aa69d07b3680b718491c5dea939e4b2f91c9e
|