AI-Parrot
AI-Parrot is an async-first Python framework for building, extending, and orchestrating AI Agents and Chatbots. Built on top of navigator-api, it provides a unified interface for interacting with various LLM providers, managing tools, conducting agent-to-agent (A2A) communication, and serving agents via the Model Context Protocol (MCP).
Whether you need a simple chatbot, a complex multi-agent orchestration workflow, or a robust production-ready AI service, AI-Parrot exposes the primitives to build it efficiently.
Monorepo Structure
AI-Parrot is organized as a monorepo with four packages:
| Package | PyPI Name | Description |
|---|---|---|
packages/ai-parrot |
ai-parrot |
Core framework: agents, clients, memory, orchestration |
packages/ai-parrot-tools |
ai-parrot-tools |
Tool and toolkit implementations (Jira, AWS, Slack, etc.) |
packages/ai-parrot-loaders |
ai-parrot-loaders |
Document loaders for RAG pipelines (PDF, YouTube, audio, etc.) |
packages/ai-parrot-pipelines |
ai-parrot-pipelines |
Specialized pipelines such as planogram compliance workflows |
The core package (ai-parrot) provides the base abstractions (AbstractTool, AbstractToolkit, @tool) and lightweight built-in tools. Heavy tool implementations, document loaders, and specialized pipelines are split into their own packages so you only install what you need.
Installation
Core framework
uv pip install ai-parrot
Quick Setup (CLI)
After installing, use the parrot CLI to configure your environment interactively:
# Interactive setup wizard — select LLM provider, enter API keys, generate .env
parrot setup
# Initialize configuration directory structure (env/ and etc/)
parrot conf init
The parrot setup wizard will guide you through:
- Selecting an LLM provider (OpenAI, Anthropic, Google, etc.)
- Entering your API credentials
- Writing them to the correct
.envfile - Optionally creating a starter Agent and bootstrap files (
app.py,run.py)
Additional CLI commands:
# Start an MCP server from a YAML config
parrot mcp --config server.yaml
# Deploy an autonomous agent as a systemd service
parrot autonomous create --agent my_agent.py
parrot autonomous install --agent my_agent.py --name my-agent
LLM Providers
Install only the providers you need:
# Google Gemini
uv pip install "ai-parrot[google]"
# OpenAI / GPT
uv pip install "ai-parrot[openai]"
# Anthropic / Claude (HTTP API client)
uv pip install "ai-parrot[anthropic]"
# Claude Code agent dispatch (bundled `claude` CLI subprocess)
uv pip install "ai-parrot[claude-agent]"
# Groq
uv pip install "ai-parrot[groq]"
# X.AI / Grok
uv pip install "ai-parrot[xai]"
# All LLM providers at once
uv pip install "ai-parrot[llms]"
Additional providers supported out of the box (no extra install needed):
- HuggingFace (
hf) — uses the HuggingFace Inference API - vLLM (
vllm) — connects to a local vLLM server - OpenRouter (
openrouter) — routes to any model via OpenRouter API - Ollama / Local — via OpenAI-compatible endpoints
Anthropic: API client vs. Claude Code agent dispatch
Anthropic ships in two independent extras — pick the one(s) you need:
| Extra | Installs | Use case |
|---|---|---|
ai-parrot[anthropic] |
anthropic[aiohttp]>=0.97.0 |
API client (AnthropicClient) — completion, vision, streaming, the Messages Batches API. Talks HTTP to api.anthropic.com. |
ai-parrot[claude-agent] |
claude-agent-sdk>=0.1.68 (which bundles the claude CLI) |
Agent dispatch (ClaudeAgentClient) — delegates a task to a Claude Code sub-agent that can read files, run bash, call tools. Talks to a subprocess CLI. |
The two extras are independent. Install only what you use:
# I just want to call the Anthropic API:
uv pip install "ai-parrot[anthropic]"
# I want to dispatch tasks to a Claude Code agent:
uv pip install "ai-parrot[claude-agent]"
# I want both:
uv pip install "ai-parrot[anthropic,claude-agent]"
After installing [claude-agent], register/authenticate the bundled CLI
once — either run claude auth interactively, or export
ANTHROPIC_API_KEY in the environment. The CLI honours either path.
A runnable demo lives in
examples/clients/claude_agent_example.py.
Embeddings & Vector Stores
# Sentence transformers, FAISS, ChromaDB, etc.
uv pip install "ai-parrot[embeddings]"
Tools
# Install the tools package
uv pip install ai-parrot-tools
# Or with specific tool extras
uv pip install "ai-parrot-tools[jira]"
uv pip install "ai-parrot-tools[aws]"
uv pip install "ai-parrot-tools[slack]"
uv pip install "ai-parrot-tools[finance]"
uv pip install "ai-parrot-tools[all]" # All tool dependencies
Available tool extras: jira, slack, aws, docker, git, analysis, excel, sandbox, codeinterpreter, pulumi, sitesearch, office365, scraping, finance, db, flowtask, google, arxiv, wikipedia, weather, messaging.
Document Loaders
# Install the loaders package
uv pip install ai-parrot-loaders
# Or with specific loader extras
uv pip install "ai-parrot-loaders[youtube]"
uv pip install "ai-parrot-loaders[pdf]"
uv pip install "ai-parrot-loaders[audio]"
uv pip install "ai-parrot-loaders[all]" # All loader dependencies
Available loader extras: youtube, audio, pdf, web, ebook, video.
Pipelines
# Install the pipelines package
uv pip install ai-parrot-pipelines
Backward-compatible imports from parrot.pipelines continue to work when the package is installed.
Platform & Security Tools
AI-Parrot includes tools for cloud security auditing and infrastructure management. These tools rely on external Docker images that must be installed before use:
# Security tools
parrot install cloudsploit # AWS security scanner (CloudSploit)
parrot install prowler # Cloud security posture management
# Platform tools
parrot install pulumi # Infrastructure as Code CLI
The parrot install command pulls and configures the required Docker containers automatically, so the tools are ready to be used by your agents.
Quick Start
Create a simple weather chatbot in just a few lines of code:
import asyncio
from parrot.bots import Chatbot
from parrot.tools import tool
# 1. Define a tool
@tool
def get_weather(location: str) -> str:
"""Get the current weather for a location."""
return f"The weather in {location} is Sunny, 25C"
async def main():
# 2. Create the Agent
bot = Chatbot(
name="WeatherBot",
llm="openai:gpt-4o", # Provider:Model
tools=[get_weather],
system_prompt="You are a helpful weather assistant."
)
# 3. Configure (loads tools, connects to memory)
await bot.configure()
# 4. Chat!
response = await bot.ask("What's the weather like in Madrid?")
print(response)
if __name__ == "__main__":
asyncio.run(main())
Using LLM Clients Directly
Beyond the Chatbot abstraction, you can access any LLM provider client directly for lower-level operations like image generation, embeddings, or custom completion calls:
import asyncio
from parrot.clients.google.client import GoogleGenAIClient
from parrot.models.outputs import ImageGenerationPrompt
from parrot.models.google import GoogleModel
async def main():
prompt = ImageGenerationPrompt(
prompt="A realistic passport-style photo with white background",
styles=["photorealistic", "high resolution"],
model=GoogleModel.IMAGEN_3.value,
aspect_ratio="16:9",
)
client = GoogleGenAIClient()
async with client:
response = await client.image_generation(prompt_data=prompt)
for img_path in response.images:
print(f"Image saved to: {img_path}")
if __name__ == "__main__":
asyncio.run(main())
Each provider client (GoogleGenAIClient, OpenAIClient, AnthropicClient, etc.) implements AbstractClient and can be used as an async context manager. This gives you full access to provider-specific features — image generation, audio transcription, structured outputs — while still benefiting from AI-Parrot's unified configuration and credential management.
Running as a Server
AI-Parrot is not only a library — it is also a full aiohttp-based application server that exposes your agents as REST APIs, WebSocket endpoints, and more. This is powered by Navigator, an async web framework built on aiohttp.
How it works
When you run parrot setup, it generates two files:
app.py— Defines your application handler, registers agents withBotManager, and configures routes.run.py— The entry point that starts the aiohttp server.
app.py (generated by parrot setup):
from parrot.manager import BotManager
from parrot.conf import STATIC_DIR
from parrot.handlers import AppHandler
from agents.my_agent import MyAgent
class Main(AppHandler):
app_name: str = "Parrot"
enable_static: bool = True
staticdir: str = STATIC_DIR
def configure(self) -> None:
self.bot_manager = BotManager()
self.bot_manager.register(MyAgent())
self.bot_manager.setup(self.app)
run.py (generated by parrot setup):
from navigator import Application
from app import Main
app = Application(Main, enable_jinja2=True)
if __name__ == "__main__":
app.run()
Built-in endpoints
Once the server starts, BotManager.setup() automatically registers these routes:
| Endpoint | Method | Description |
|---|---|---|
/api/v1/agents/chat/{agent_id} |
POST | Chat with an agent (JSON, HTML, or Markdown response) |
/api/v1/agents/chat/{agent_id} |
PATCH | Configure tools/MCP servers for a session |
/api/v1/bot_management |
GET | List registered bots |
/api/v1/bot_management/{bot} |
GET/POST/PATCH/DELETE | CRUD operations on bots |
/api/v1/agent_tools |
GET | List available tools |
/api/v1/ai/client |
GET | LLM provider configuration |
/ws/userinfo |
WebSocket | Real-time user notifications |
Starting the server
Development (single process, auto-reload):
python run.py
The server starts on http://0.0.0.0:5000 by default (configurable via APP_HOST / APP_PORT environment variables).
Production (Gunicorn with async workers):
# Install gunicorn
uv pip install "ai-parrot[deploy]"
# Run with aiohttp-compatible workers
gunicorn run:app \
--worker-class aiohttp.worker.GunicornUVLoopWebWorker \
--workers 4 \
--bind 0.0.0.0:5000 \
--timeout 360
The long timeout (360s) accommodates agent queries that involve multi-step tool execution or LLM calls.
Talking to your agents via REST
Once the server is running, any registered agent is accessible via HTTP:
# Chat with an agent
curl -X POST http://localhost:5000/api/v1/agents/chat/my-agent \
-H "Content-Type: application/json" \
-d '{"message": "What is the weather in Madrid?"}'
# Request markdown output
curl -X POST "http://localhost:5000/api/v1/agents/chat/my-agent?output_format=markdown" \
-H "Content-Type: application/json" \
-d '{"message": "Summarize the latest news"}'
Architecture
AI-Parrot is designed with a modular architecture enabling agents to be both consumers and providers of tools and services.
graph TD
User["User / Client"] --> API["AgentTalk Handlers"]
API --> Bot["Chatbot / BaseBot"]
subgraph "Agent Core"
Bot --> Memory["Memory / Vector Store"]
Bot --> LLM["LLM Client (OpenAI/Anthropic/Etc)"]
Bot --> TM["Tool Manager"]
end
subgraph "Tools & Capabilities"
TM --> LocalTools["Local Tools (@tool)"]
TM --> Toolkits["Toolkits (OpenAPI/Custom)"]
TM --> MCPServer["External MCP Servers"]
end
subgraph "Connectivity"
Bot -.-> A2A["A2A Protocol (Client/Server)"]
Bot -.-> MCP["MCP Protocol (Server)"]
Bot -.-> Integrations["Telegram / MS Teams"]
end
subgraph "Orchestration"
Crew["AgentCrew"] --> Bot
Crew --> OtherBots["Other Agents"]
end
Core Concepts
Agents (Chatbot)
The Chatbot class is your main entry point. It handles conversation history, RAG (Retrieval-Augmented Generation), and the tool execution loop.
bot = Chatbot(
name="MyAgent",
model="anthropic:claude-3-5-sonnet-20240620",
enable_memory=True
)
Tools
Functional Tools (@tool)
The simplest way to create a tool. The docstring and type hints are automatically used to generate the schema for the LLM.
from parrot.tools import tool
@tool
def calculate_vat(amount: float, rate: float = 0.20) -> float:
"""Calculate VAT for a given amount."""
return amount * rate
Class-Based Toolkits (AbstractToolkit)
Group related tools into a reusable class. All public async methods become tools.
from parrot.tools import AbstractToolkit
class MathToolkit(AbstractToolkit):
async def add(self, a: int, b: int) -> int:
"""Add two numbers."""
return a + b
async def multiply(self, a: int, b: int) -> int:
"""Multiply two numbers."""
return a * b
OpenAPI Toolkit (OpenAPIToolkit)
Dynamically generate tools from any OpenAPI/Swagger specification.
from parrot.tools import OpenAPIToolkit
petstore = OpenAPIToolkit(
spec="https://petstore.swagger.io/v2/swagger.json",
service="petstore"
)
# Now your agent can call petstore_get_pet_by_id, etc.
bot = Chatbot(name="PetBot", tools=petstore.get_tools())
Orchestration (AgentCrew)
Orchestrate multiple agents to solve complex tasks using AgentCrew.
Supported Modes:
- Sequential: Agents run one after another, passing context.
- Parallel: Independent tasks run concurrently.
- Flow: DAG-based execution defined by dependencies.
- Loop: Iterative execution until a condition is met.
from parrot.bots.orchestration import AgentCrew
crew = AgentCrew(
name="ResearchTeam",
agents=[researcher_agent, writer_agent]
)
# Define a Flow — Writer waits for Researcher to finish
crew.task_flow(researcher_agent, writer_agent)
await crew.run_flow("Research the latest advancements in Quantum Computing")
Scheduling (@schedule)
Give your agents agency to run tasks in the background.
from parrot.scheduler import schedule, ScheduleType
class DailyBot(Chatbot):
@schedule(schedule_type=ScheduleType.DAILY, hour=9, minute=0)
async def morning_briefing(self):
news = await self.ask("Summarize today's top tech news")
await self.send_notification(news)
Connectivity & Exposure
Agent-to-Agent (A2A) Protocol
Agents can discover and talk to each other using the A2A protocol.
Expose an Agent:
from parrot.a2a import A2AServer
a2a = A2AServer(my_agent)
a2a.setup(app, url="https://my-agent.com")
Consume an Agent:
from parrot.a2a import A2AClient
async with A2AClient("https://remote-agent.com") as client:
response = await client.send_message("Hello from another agent!")
Model Context Protocol (MCP)
AI-Parrot has first-class support for MCP.
Consume MCP Servers:
mcp_servers = [
MCPServerConfig(
name="filesystem",
command="npx",
args=["-y", "@modelcontextprotocol/server-filesystem", "/home/user"]
)
]
await bot.setup_mcp_servers(mcp_servers)
Expose Agent as MCP Server: Allow Claude Desktop or other MCP clients to use your agent as a tool.
Platform Integrations
Expose your bots natively to chat platforms:
- Telegram
- Microsoft Teams
- Slack
Optional capabilities
Dev-Loop Orchestration
Optional. Requires the
[claude-agent]extra:pip install ai-parrot[claude-agent]
A 5-node AgentsFlow that fixes "small operational bugs" automatically:
BugIntake → Research → Development → QA → DeploymentHandoff
│
└─(qa failed / hard error)→ FailureHandler
The flow takes a Pydantic BugBrief (Jira ticket + log sources +
acceptance criteria) and produces a PR plus a Jira ticket transitioned
to "Ready to Deploy". Failures escalate back to the original reporter.
Prerequisites
- Python 3.11+ with
ai-parrot[claude-agent]installed. claude-agent-sdk >= 0.1.68and eitherANTHROPIC_API_KEYor a configuredclaudeCLI onPATH.- Redis 6+ for two-stream observability (one stream per flow run plus one per dispatch).
- Jira service-account credentials wrapped in a
parrot.auth.credentials.StaticCredentialResolver. - (Optional)
ghCLI for PR creation. Falls back to a direct GitHub REST call (usingGITHUB_TOKEN+GITHUB_REPOSITORY) when the CLI is missing.
Configuration (navconfig)
| Setting | Default | Purpose |
|---|---|---|
CLAUDE_CODE_MAX_CONCURRENT_DISPATCHES |
3 |
Cap on concurrent Claude Code dispatches (dispatcher-side semaphore). |
FLOW_MAX_CONCURRENT_RUNS |
5 |
Cap on concurrent flow runs (orchestrator-side). |
FLOW_BOT_JIRA_ACCOUNT_ID |
"" |
Jira accountId of the service-account bot. Must be set per environment. |
WORKTREE_BASE_PATH |
.claude/worktrees |
Base directory for per-feature worktrees. The dispatcher refuses any cwd outside this path. |
FLOW_STREAM_TTL_SECONDS |
604800 |
Retention for both flow and dispatch Redis streams (7 days). |
ACCEPTANCE_CRITERION_ALLOWLIST |
["flowtask","pytest","ruff","mypy","pylint"] |
Allowed ShellCriterion command heads. Validated at intake. |
Quickstart
from parrot.flows.dev_loop import (
ClaudeCodeDispatcher,
build_dev_loop_flow,
register_pull_request_webhook,
)
dispatcher = ClaudeCodeDispatcher(
max_concurrent=3,
redis_url="redis://localhost:6379/0",
stream_ttl_seconds=604800,
)
flow = build_dev_loop_flow(
dispatcher=dispatcher,
jira_toolkit=jira, # already wrapping flow-bot creds
log_toolkits={"cloudwatch": cw, "elasticsearch": es},
redis_url="redis://localhost:6379/0",
)
register_pull_request_webhook(orchestrator, secret=GITHUB_WEBHOOK_SECRET)
# Then run via your AutonomousOrchestrator with a BugBrief in ctx.
Live observability
The dispatcher publishes per-event DispatchEvent envelopes (queued,
started, message, tool_use, tool_result, output_invalid,
failed, completed) to Redis Streams. The
parrot.flows.dev_loop.flow_stream_ws aiohttp handler exposes a
WebSocket endpoint that fans-in the flow stream and every dispatch
stream into a single envelope per event for the UI to consume — the UI
never speaks Redis directly.
Supported LLM Providers
| Provider | Extra | Identifier | Example |
|---|---|---|---|
| OpenAI | openai |
openai |
openai:gpt-4o |
| Anthropic | anthropic |
anthropic, claude |
anthropic:claude-sonnet-4-20250514 |
| Google Gemini | google |
google |
google:gemini-3.1-flash-lite-preview |
| Groq | groq |
groq |
groq:llama-3.3-70b-versatile |
| X.AI / Grok | xai |
grok |
grok:grok-3 |
| HuggingFace | (included) | hf |
hf:meta-llama/Llama-3-8B |
| vLLM | (included) | vllm |
vllm:model-name |
| OpenRouter | (included) | openrouter |
openrouter:anthropic/claude-sonnet-4 |
| Ollama | (included) | via OpenAI endpoint | — |
Contributing
Development setup (from source)
AI-Parrot uses uv as its package manager and provides a Makefile to simplify common tasks.
git clone https://github.com/phenobarbital/ai-parrot.git
cd ai-parrot
# Create the virtual environment (Python 3.11)
make venv
source .venv/bin/activate
# Full dev install — all packages, all extras, dev tools
make develop
# Run tests
make test
Makefile targets
The Makefile covers the entire development lifecycle. Run make help for the full list.
Development install variants:
| Target | What it installs |
|---|---|
make develop |
All packages + all extras + dev tools (full environment) |
make develop-fast |
All packages, base deps only (no torch/tensorflow/whisperx) |
make develop-ml |
Embeddings + audio loaders (heavy ML stack) |
Production install variants:
| Target | What it installs |
|---|---|
make install |
All packages, base deps only (no extras) |
make install-core |
Core with LLM clients + vector stores |
make install-tools |
Core + tools with common extras (jira, slack, aws, etc.) |
make install-tools-all |
Core + tools with ALL extras |
make install-loaders |
Core + loaders with common extras (youtube, web, pdf) |
make install-loaders-all |
Core + loaders with ALL extras (includes whisperx, pyannote) |
make install-all |
Everything with ALL extras |
Other useful targets:
make format # Format code with black
make lint # Lint with pylint + black --check
make test # Run pytest + mypy
make build # Build all packages (sdist + wheel)
make release # Build + publish to PyPI
make lock # Regenerate uv.lock
make clean # Remove build artifacts
make generate-registry # Regenerate TOOL_REGISTRY from source
make bump-patch # Bump patch version (syncs across all packages)
Manual install (without Make)
If you prefer not to use Make:
uv venv --python 3.11 .venv
source .venv/bin/activate
# Full install
uv sync --all-packages --all-extras
# Or selective extras
uv sync --extra google --extra openai
Project layout
ai-parrot/
├── packages/
│ ├── ai-parrot/ # Core framework
│ │ └── src/parrot/
│ ├── ai-parrot-tools/ # Tool implementations
│ │ └── src/parrot_tools/
│ └── ai-parrot-loaders/ # Document loaders
│ └── src/parrot_loaders/
├── tests/
├── examples/
├── Makefile # Build, install, test, release shortcuts
└── pyproject.toml # Workspace root
Releasing to PyPI
AI-Parrot publishes three packages on every GitHub release:
| Package | PyPI Project | Build Method |
|---|---|---|
ai-parrot |
ai-parrot | cibuildwheel (Cython + Rust/Maturin) |
ai-parrot-tools |
ai-parrot-tools | uv build (pure Python) |
ai-parrot-loaders |
ai-parrot-loaders | uv build (pure Python) |
The release workflow (.github/workflows/release.yml) runs 3 parallel build jobs and a single deploy job:
release event
├── build-core — cibuildwheel for ai-parrot (Cython + Rust)
├── build-tools — uv build for ai-parrot-tools
├── build-loaders — uv build for ai-parrot-loaders
└── deploy — twine upload all artifacts to PyPI
To create a release:
- Bump the version in each package's
pyproject.toml(or usemake bump-patchto sync all three). - Create a GitHub release — the workflow triggers automatically on the
release: createdevent.
First-time PyPI setup (required once):
- Create
ai-parrot-toolsandai-parrot-loadersprojects on PyPI under the same account asai-parrot. - Ensure the
NAV_AIPARROT_API_SECRETGitHub secret holds a PyPI API token with upload scope for all 3 projects. A scoped token per project or a single account-level token both work.
Independent versioning:
Each package has its own version number in its pyproject.toml. All three are built and published on the same release event — there is no requirement to keep versions in sync.
Guidelines
- All code must be async-first — no blocking I/O in async contexts
- Use type hints and Google-style docstrings on all public APIs
- Use Pydantic models for structured data
- Run
pytestafter any logic change - Tools with heavy dependencies must use lazy imports to avoid bloating the core
Issues & Support
- Issues: GitHub Tracker
- Discussion: GitHub Discussions
DB-Persisted Bot Configuration (FEAT-133)
Added in FEAT-133. Spec:
sdd/specs/bot-reranker-and-parent-searcher-config.spec.md.
BotManager._load_database_bots reads two new JSONB columns on
navigator.ai_bots and wires the resulting objects into every DB-loaded bot
at startup. Both columns default to '{}'::JSONB, so existing rows are
unaffected.
Reranker config (reranker_config)
Controls cross-encoder or LLM-based reranking of vector-store candidates
(FEAT-126). Empty {} means no reranking.
Local cross-encoder (no live LLM call; requires sentence-transformers):
{
"type": "local_cross_encoder",
"model_name": "cross-encoder/ms-marco-MiniLM-L-12-v2",
"device": "cpu",
"rerank_oversample_factor": 4
}
LLM reranker (uses the bot's own LLM client):
{
"type": "llm",
"client_ref": "bot",
"rerank_oversample_factor": 4
}
Factory: parrot.rerankers.factory.create_reranker(config, *, bot_llm_client=None)
Parent-searcher config (parent_searcher_config)
Controls parent-document expansion after vector search (FEAT-128).
Empty {} means no expansion.
In-table parent search (chunk row has a parent_id FK in the same table):
{
"type": "in_table",
"expand_to_parent": true
}
expand_to_parent is also forwarded as a constructor kwarg so the bot's
retrieval logic can branch on it before calling the searcher.
Factory: parrot.stores.parents.factory.create_parent_searcher(config, *, store)
store(bot.store) becomes available only afterawait bot.configure(app). The factory is therefore called afterconfigure(), not before.
Error handling
An unknown type value raises parrot.exceptions.ConfigError at bot startup.
The bot is not silently registered without its configured features.
# A row with {"type": "magic"} in reranker_config will raise:
# ConfigError: unknown reranker type 'magic'; supported: local_cross_encoder, llm
Order of operations
create_reranker(reranker_config) # before bot construction
bot = BotClass(..., reranker=reranker, expand_to_parent=...)
await bot.configure(app) # store becomes available
create_parent_searcher(config, store=bot.store)
bot.parent_searcher = parent_searcher
License
MIT
Built with care by the AI-Parrot Team
1 maintainer
Metadata
1 maintainer
Release files for ai-parrot 1.0.3
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| ai_parrot-1.0.3.tar.gz | 5.0 MB | Details |
Built distributions (wheels)
| File | Reset | |||
|---|---|---|---|---|
| ai_parrot-1.0.3-cp313-cp313-musllinux_1_2_x86_64.whl | CPython 3.13 | CPython 3.13 | Linux musl 1.2+ x86-64 | Details |
| ai_parrot-1.0.3-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl | CPython 3.13 | CPython 3.13 | Linux glibc 2.28+ x86-64, Linux glibc 2.17+ x86-64 | Details |
| ai_parrot-1.0.3-cp312-cp312-musllinux_1_2_x86_64.whl | CPython 3.12 | CPython 3.12 | Linux musl 1.2+ x86-64 | Details |
| ai_parrot-1.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl | CPython 3.12 | CPython 3.12 | Linux glibc 2.17+ x86-64, Linux glibc 2.28+ x86-64 | Details |
| ai_parrot-1.0.3-cp311-cp311-musllinux_1_2_x86_64.whl | CPython 3.11 | CPython 3.11 | Linux musl 1.2+ x86-64 | Details |
| ai_parrot-1.0.3-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl | CPython 3.11 | CPython 3.11 | Linux glibc 2.28+ x86-64, Linux glibc 2.17+ x86-64 | Details |
Total release size: 40.6 MB
Release files / ai_parrot-1.0.3.tar.gz
| Download URL | ai_parrot-1.0.3.tar.gz |
|---|---|
| Size | 5.0 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
033497392d4a0d43cb6c0a0686ea6ec24117bdef2345f860bbf16eb05e537ee9
|
|
BLAKE2b-256 checksum How to use checksums |
075d80631d0250b755fc2858d122cb548954ce979beb7292b9bcc114e412be27
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 17, 2026.
Transparency logRelease files / ai_parrot-1.0.3-cp313-cp313-musllinux_1_2_x86_64.whl
| Download URL | ai_parrot-1.0.3-cp313-cp313-musllinux_1_2_x86_64.whl |
|---|---|
| Size | 5.9 MB |
| Tags | CPython 3.13 Linux musl 1.2+ x86-64 |
|
SHA-256 checksum How to use checksums |
0d7b2e03649b4dec3d438c9b82c3ccbf9b912634b64bb08963af749e219fb8f1
|
|
BLAKE2b-256 checksum How to use checksums |
c4fa78421c87889b5544227f5b475bdd6213124bdfd13f70b9b8084f4e92f2a5
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 17, 2026.
Transparency logRelease files / ai_parrot-1.0.3-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl
| Download URL | ai_parrot-1.0.3-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl |
|---|---|
| Size | 5.9 MB |
| Tags | CPython 3.13 Linux glibc 2.17+ x86-64 Linux glibc 2.28+ x86-64 |
|
SHA-256 checksum How to use checksums |
24b497ee8133949aab92f7857294f46d4dc5180cb22fa926f086f797a93a83c9
|
|
BLAKE2b-256 checksum How to use checksums |
a5da34a3cb7227b6ef219d4a8bac3f100d36322a9c7c45fddf3bb125582c63a2
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 17, 2026.
Transparency logRelease files / ai_parrot-1.0.3-cp312-cp312-musllinux_1_2_x86_64.whl
| Download URL | ai_parrot-1.0.3-cp312-cp312-musllinux_1_2_x86_64.whl |
|---|---|
| Size | 5.9 MB |
| Tags | CPython 3.12 Linux musl 1.2+ x86-64 |
|
SHA-256 checksum How to use checksums |
7ecf971a73855619cea340d036063a979bee0bc50253f01d59dbe921f965eb23
|
|
BLAKE2b-256 checksum How to use checksums |
ffe82293d65c607be4a1bea70437e559bc78adec3511e1d8845d92172cc031ac
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 17, 2026.
Transparency logRelease files / ai_parrot-1.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl
| Download URL | ai_parrot-1.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl |
|---|---|
| Size | 6.0 MB |
| Tags | CPython 3.12 Linux glibc 2.17+ x86-64 Linux glibc 2.28+ x86-64 |
|
SHA-256 checksum How to use checksums |
4cc7a5a948928cd66c37bd7e59ffb86595ff5a0c212741ac23fe0f6e6da310e5
|
|
BLAKE2b-256 checksum How to use checksums |
d7335fbe20665e1953084564e02a42d135441f86c37a2bcfc5e8d2414a2247d5
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 17, 2026.
Transparency logRelease files / ai_parrot-1.0.3-cp311-cp311-musllinux_1_2_x86_64.whl
| Download URL | ai_parrot-1.0.3-cp311-cp311-musllinux_1_2_x86_64.whl |
|---|---|
| Size | 5.9 MB |
| Tags | CPython 3.11 Linux musl 1.2+ x86-64 |
|
SHA-256 checksum How to use checksums |
a71ca51dc566d779ae81f6c51ef52e6f03c48aad10fa61a6bfc2569ef892b943
|
|
BLAKE2b-256 checksum How to use checksums |
1809bbf587402b1eb044d509429b2ac1921b8d2e6fa48038c32cb41923495f02
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 17, 2026.
Transparency logRelease files / ai_parrot-1.0.3-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl
| Download URL | ai_parrot-1.0.3-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl |
|---|---|
| Size | 5.9 MB |
| Tags | CPython 3.11 Linux glibc 2.17+ x86-64 Linux glibc 2.28+ x86-64 |
|
SHA-256 checksum How to use checksums |
d600634c41a06260355aef7994c6c74dbf03677c07a09fa95d94ff4ca1f2c290
|
|
BLAKE2b-256 checksum How to use checksums |
86fa213f0cd8893a694638136c40f766c0afb92e6b66dd88a7240cd6bf2e49df
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 17, 2026.
Transparency log
