AI Bridge Interface & Gateway Abstraction Layer

Abigail is a powerful Command-Line Interface (CLI) tool and Python library for accessing free web AI services (chat models, image & video generation) directly from your terminal or Python code-with zero API keys required.

Language: English · Bahasa Indonesia

Caution

Disclaimer: This package accesses third-party public web AI interfaces. Please use it responsibly and in accordance with each service's Terms of Service.

---

Quick Start

Installation

Requires Python ≥ 3.12.

Install as CLI tool:

pip install abigail
# or with uv:
uv tool install abigail

Install as dependency in your Python project:

pip install abigail
# or with uv:
uv add abigail

Install from source (for development):

git clone https://github.com/iqbalmh18/abigail.git
cd abigail
uv sync
source .venv/bin/activate

Dependencies

All dependencies are installed automatically with pip install abigail or uv add abigail.

Package	Version	Purpose
pydantic	≥ 2.13.4	Request/response models and config validation.
typer	≥ 0.14.0	CLI framework.
curl-cffi	≥ 0.15.0	HTTP client with TLS fingerprint spoofing.
tomli-w	≥ 1.2.0	Writing TOML config files.
quickjs	≥ 1.19.4	Embedded JavaScript runtime for challenge solving.
nodriver	≥ 0.50.3	Headless browser automation (required for ImageClient).
term-image	≥ 0.7.2	Rendering images inline in the terminal.
rich	≥ 15.0.0	Terminal formatting and output rendering.
pillow	≥ 10.4.0	Image processing.

Note

nodriver is required for ImageClient (image generation) and VideoClient (video generation). It drives a real Chromium browser in the background to solve browser challenges. On headless servers, make sure Chromium and its dependencies are available, or set headless=false / abigail config set headless false if you need a visible browser window for debugging.

---

CLI Usage

Abigail features a rich command-line tool (abigail) so you can stream AI responses, generate images, or pipe terminal data directly into LLMs without writing code.

Quick Commands Overview

# 1. Ask a question & stream response to stdout
abigail chat "Explain quantum computing in simple terms"

# 2. Generate an image
abigail imagen "a cute sleeping cat on a cozy rug"

# 3. Generate a video
abigail vidgen "a cat wearing a suit dancing in Times Square"

# 4. Enter interactive terminal chat mode
abigail shell

All Available CLI Commands

Command	Description
abigail chat PROMPT	Send a prompt to an AI model and stream the response to terminal.
abigail imagen PROMPT	Generate images using AI providers (e.g., Perchance).
abigail vidgen PROMPT	Generate videos using AI providers (e.g., Upsampler), with optional image-to-video.
abigail shell	Start an interactive chat session (/help, /exit).
abigail providers	List all registered AI providers and their capabilities.
abigail models	List all supported AI models and aliases.
abigail inspect <provider>	Display provider metadata (capabilities, models, aliases).
abigail benchmark [PROMPT]	Benchmark providers & models with success rate and latency stats.
abigail config get|set|unset|reset	View or manage local CLI config (~/.config/abigail/config.toml).

---

Supported Providers & Models

Capabilities are listed per model (✓ = supported, - = not supported).

Provider	Alias	Capability	Thinking	Search	Priority
deepai	deepai/default	Chat	-	-	100
deepai	deepai/deepseek-v3.2	Chat	-	-	100
deepai	deepai/llama-4-scout	Chat	-	-	100
deepai	deepai/gpt-oss-120b	Chat	✓	-	100
deepai	deepai/gemini-2.5-flash-lite	Chat	-	-	100
deepai	deepai/gemma-4	Chat	-	-	100
deepai	deepai/gpt-4.1-nano	Chat	-	-	100
deepai	deepai/gpt-5-nano	Chat	-	-	100
deepai	deepai/llama-3.3-70b-instruct	Chat	-	-	100
deepai	deepai/llama-3.1-8b-instant	Chat	-	-	100
heckai	heckai/gpt-5.4-mini	Chat	-	-	100
heckai	heckai/gemini-3.1-flash-preview	Chat	-	-	100
heckai	heckai/gemini-3-flash-lite	Chat	-	-	100
heckai	heckai/deepseek-v4-flash	Chat	✓	✓	100
heckai	heckai/deepseek-v4-pro	Chat	✓	✓	100
heckai	heckai/hy3-preview	Chat	✓	✓	100
heckai	heckai/qwen3.7-plus	Chat	✓	✓	100
heckai	heckai/step-3.7-flash	Chat	✓	✓	100
notrack	notrack/default	Chat	✓	-	100
notrack	notrack/minimax	Chat	✓	-	100
notrack	notrack/chatgpt	Chat	✓	-	100
perchance	perchance/text-to-image	Chat	-	-	100
quillbot	quillbot/default	Chat	-	-	100
unlimitedai	unlimitedai/default	Image	-	-	100
upsampler	upsampler/ltx-video	Video	-	-	100

---

CLI Examples & Tips

chat Examples

# Stream response using a specific provider & model
abigail chat "Explain quantum computing" -p heckai -m heckai/gpt-5.4-mini

# Use the notrack provider to upload & analyze files or PDFs
abigail chat "Describe this image" -p notrack -f /path/to/image.jpg
abigail chat "Summarize this PDF" -p notrack -f sample.pdf

# Non-streaming JSON output
abigail chat "Hello" --no-stream --json

# Enable thinking/reasoning and live web search
abigail chat "Latest AI news" --thinking --search --timeout 60 --proxy socks5://host:port

Note

Auto file routing: When passing -f/--files with provider="auto", Abigail automatically picks a provider that supports file attachments (like notrack).

Tip

Rate limited? If chat, imagen, or vidgen hits a rate limit (429), route the request through a proxy with the --proxy argument, e.g. abigail chat "Hi" --proxy socks5://host:port or abigail imagen "a cat" --proxy http://127.0.0.1:8080. You can also set a default proxy or a rotating proxy pool via abigail config set proxy proxies.txt.

Unix Pipelines (stdin)

abigail chat, abigail imagen, and abigail vidgen natively accept piped data from standard input (stdin):

# Pipe code into Abigail for review
cat main.py | abigail chat "Review this file in depth"

# Pipe git diff to write a commit message
git diff | abigail chat "Write a concise commit message for this diff"

# Chain LLM prompt output into image generation
abigail chat "Write a detailed prompt for a fantasy castle" | abigail imagen -r 1024x1024

# Stdin only (no prompt argument needed)
echo "Explain what a monad is" | abigail chat

imagen Examples

# Generate image with default resolution (512x768)
abigail imagen "a cute sleeping cat"

# Custom resolution, guidance scale, and output path
abigail imagen "a futuristic cyberpunk city" -r 1024x1024 -g 9.0 -o city.jpeg

# Pipe prompt via stdin
echo "a majestic dragon flying over mountains" | abigail imagen -r 768x768

# Disable headless mode (shows browser window, useful for debugging)
abigail imagen "a cute cat" --no-headless

vidgen Examples

# Generate a video with default settings (768x512, 3 seconds)
abigail vidgen "a cat wearing a suit dancing in Times Square"

# Custom duration, resolution preset, and output path
abigail vidgen "a futuristic cyberpunk city" -d 5 -r 16:9 -o city.mp4

# Image-to-video: animate an existing image
abigail vidgen "a still life comes alive" -i input.jpg

# Fixed seed with prompt enhancement
abigail vidgen "a majestic dragon" -s 42 --no-randomize --enhance

# Pipe prompt via stdin
echo "waves crashing on a beach at sunset" | abigail vidgen -r portrait

Tip

-r/--resolution accepts WIDTHxHEIGHT (e.g. 768x512) or presets: landscape, portrait, square, landscape-hd, portrait-hd, square-hd, and ratios 16:9, 9:16, 1:1, 3:2, 2:3.

Configuration (abigail config)

CLI preferences are saved in ~/.config/abigail/config.toml and applied automatically:

# Set default provider & model
abigail config set provider notrack
abigail config set model notrack/ChatGPT

# General options
abigail config set timeout 120.0
abigail config set stream true
abigail config set verbose true

# Headless mode for browser-based providers (default: true)
abigail config set headless false

# View or reset all config
abigail config get
abigail config reset

All available config keys:

Key	Default	Description
provider	(not set)	Default provider (auto if unset).
model	(not set)	Default model (auto if unset).
proxy	(not set)	Proxy URL or @path/to/proxies.txt.
timeout	(not set)	Request timeout in seconds (no limit if unset).
stream	true	Stream responses by default.
thinking	false	Enable chain-of-thought reasoning by default.
search	false	Enable web search by default.
verbose	false	Show metadata and debug info by default.
failover	(not set)	true / false / "provider1,provider2" — see below.
headless	true	Run browser-based providers in headless mode.
resolution	512x768	Default image resolution for imagen.
guidance_scale	7.0	Default guidance scale for imagen.
seed	-1	Default seed for imagen (-1 = random).
negative_prompt	(empty)	Default negative prompt for imagen.

To reset a single key back to its default:

abigail config unset provider
abigail config unset headless
abigail config unset negative_prompt

Proxy & Failover Settings

# Set a single proxy or proxy pool file
abigail config set proxy "http://127.0.0.1:8080"
abigail config set proxy proxies.txt

# Failover: restrict & order the failover chain
abigail config set failover "quillbot,heckai"   # explicit order
abigail config set failover true                 # auto failover to all
abigail config set failover false                # never fail over

# Override per command
abigail chat "Hi" --no-failover
abigail chat "Hi" --failover-order quillbot,heckai

When a specific --provider is set, Abigail automatically boosts that provider's score and penalises others so it stays at the top of the failover chain.

---

Python Library Usage

Abigail is also a fully typed, asynchronous Python library (asyncio).

1. Stream Chat (Recommended)

import asyncio
from abigail import ChatClient

async def main():
    async with ChatClient() as client:
        async for chunk in client.stream_chat("Explain machine learning"):
            if chunk.event == "content":
                print(chunk.text, end="", flush=True)

asyncio.run(main())

2. Non-Streaming Chat

import asyncio
from abigail import ChatClient

async def main():
    async with ChatClient() as client:
        response = await client.chat("What is Python?")
        print(response.text)
        print("thinking:", response.thinking)

asyncio.run(main())

3. Image Generation (ImageClient)

import asyncio
from abigail import ImageClient

async def main():
    async with ImageClient() as client:
        res = await client.generate_image(
            prompt="a cute sleeping cat",
            resolution="512x768",
            guidance_scale=7.0,
        )
        img = res.images[0]
        with open("cat.jpeg", "wb") as f:
            f.write(img.image_bytes)
        print(f"Saved image from provider: {res.provider}, seed: {img.seed}")

asyncio.run(main())

4. Video Generation (VideoClient)

import asyncio
from abigail import VideoClient

async def main():
    async with VideoClient() as client:
        res = await client.generate_video(
            prompt="a cat wearing a suit dancing in Times Square",
            duration=3.0,
            width=768,
            height=512,
        )
        vid = res.videos[0]
        with open(f"cat.{vid.file_extension}", "wb") as f:
            f.write(vid.video_bytes)
        print(f"Saved video from provider: {res.provider}, seed: {vid.seed}")

asyncio.run(main())

Image-to-video is supported via the image parameter:

await client.generate_video("a still life comes alive", image="/path/to/input.jpg")

5. Advanced Capabilities: Thinking & Web Search

# Enable CoT (Chain of Thought) reasoning
await client.chat("Solve: if 2x + 3 = 11, what is x?", thinking=True)

# Enable web search / live internet access
await client.chat("Latest tech news?", search=True)

Parsing SSE event stream with thinking & search:

from abigail import ChatEventType

async for chunk in client.stream_chat("Latest AI news", thinking=True, search=True):
    match chunk.event:
        case ChatEventType.thinking:
            print(chunk.thinking, end="")
        case ChatEventType.content:
            print(chunk.text, end="")
        case ChatEventType.sources:
            for s in chunk.search:
                print(s.title, s.url)

6. Specific Provider / Model Selection

await client.chat("Hello!", provider="quillbot")
await client.chat("Explain quantum computing", provider="heckai", model="heckai/openai/gpt-5.4-mini")

7. File Attachments

# Pass local file paths directly
await client.chat("Describe this image", files=["/path/to/image.jpg"])

# Or use the Attachment model for raw bytes
from abigail.types.requests import Attachment

my_file = Attachment(name="data.csv", data=b"a,b,c\n1,2,3")
await client.chat("Analyze this data", files=[my_file])

8. Built-in Resilience & Failover

ChatClient automatically retries transient errors (429, 5xx, connection resets) and fails over across providers when provider="auto" is used. Circuit breaker temporary skips unhealthy providers automatically.

---

Architecture & Internals (For Contributors)

This section explains internal mechanics for developers interested in contributing or adding new providers.

Request Flow & Failover Logic

flowchart TD
    A["_failover_chain(provider, model, thinking, search)"] --> B{"provider == auto?"}
    B -->|yes| C["providers by priority, filtered by circuit breaker + capabilities"]
    B -->|explicit| D["single provider, no failover"]
    C --> E["for each provider: resolve model"]
    D --> E["round-robin provider models"]
    E --> F["provider._select_models() (per-provider)"]
    F --> G["filter by supports_thinking / supports_search"]
    G --> H["rotate start by next(_rr_counter)"]
    H --> I["loop POST each candidate model"]
    I -->|429 / 5xx| J["next model"]
    J --> I
    I -->|200| K["stream + parse"]
    I -->|all models failed| L["RetryableError"]
    L -->|retries left| I
    L -->|retries exhausted| M{"provider == auto?"}
    M -->|yes| N["fail over to next provider"]
    N --> E
    M -->|explicit| O["raise last error"]
    L -->|all providers exhausted| O

Loading

Adding a New Provider

To add a provider, create a package under abigail/providers/<name>/ (metadata.py, session.py, parser.py, provider.py) and register it using @register_provider. Both ChatClient and the CLI will auto-discover it without requiring edits to CLI core files.

Running Tests

uv run pytest tests/ -v
uv run pytest tests/ --cov=abigail --cov-report=term-missing

---

License

MIT License.