The Redbird Agent SDK lets developers access and orchestrate Redbird's full library of AI agents programmatically — triggering connector, data collection, transformation, analysis, and reporting workflows directly from Python, without touching the Redbird UI.
from redbird_agent_sdk import RedbirdAgentSDK import time # Initialize with your Redbird credentials sdk = RedbirdAgentSDK( user_id="your_user_id", project_folder_id="your_project_id" ) # Browse all available agents for your project sdk.get_agent_list() # Get an agent by name agent = sdk.get_agent_by_name('Autotagger Agent') # Set the required inputs and run agent.requirements['input_dataset'] = '/path/to/data.csv' agent.run() # Poll until complete, then access results while not agent.retrieve_results(): time.sleep(10) df = agent.output_objects[0].get_dataframe() print(df.head())
Redbird's AI agents handle the hardest parts of data work: pulling data from any connected source, cleaning and transforming it, running analysis, generating reporting outputs, and taking actions in source systems. The Agent SDK exposes those same agents through a simple Python interface — so engineering teams can invoke them programmatically, chain them together, and integrate their outputs into any downstream system or application.
The full Redbird agent library is available through the SDK — Connector agents, Data Transformation agents, Analysis agents, and more.
Initialize the SDK, browse available agents, configure inputs, and trigger execution — all from Python code in any environment.
Agent outputs are returned as structured Python objects — pipe them into your own data pipelines, databases, applications, or orchestration layers.
The SDK gives you the same AI agent capabilities available in the Redbird UI — with the precise control and composability that code provides.
Jupyter notebooks, scripts, data pipelines, scheduled jobs, CI/CD workflows — if it runs Python, it can use the SDK.
Connect with your Redbird user ID and project folder ID
RedbirdAgentSDK(user_id, project_folder_id)
List available agents, get one by ID or name, inspect its required inputs
sdk.get_agent_by_name('...')
Pass inputs, invoke the agent, poll for completion, and handle structured outputs
agent.run() → get_dataframe()
Every AI agent available in the Redbird platform is accessible through the SDK. Once initialized, you can retrieve the full list of available agents for your project — each with its name, description, and the inputs it requires. Browse, select, and invoke any agent from your code.
from redbird_agent_sdk import RedbirdAgentSDK # Initialize with your Redbird credentials sdk = RedbirdAgentSDK( user_id="<<your_user_id>>", project_folder_id="<<your_project_id>>" ) # Load the available agent list for your project agents = sdk.get_agent_list() # Each agent has name, description, and required inputs for a in agents: print(a['name'], '—', a['description'])
[ { 'id': 1, 'name': 'Connector Agent', 'description': 'Pull data from any connected source', 'requirements': ['source', 'query'] }, { 'id': 7, 'name': 'Data Transformation Agent', 'description': 'Clean, join, and enrich datasets', 'requirements': ['input_dataset', 'instructions'] }, { 'id': 19, 'name': 'Autotagger Agent', 'description': 'Classify and tag records automatically', 'requirements': ['input_dataset', 'tag_column'] }, ... ]
import time from redbird_agent_sdk import RedbirdAgentSDK sdk = RedbirdAgentSDK(user_id, project_folder_id) # Retrieve an agent and inspect its requirements agent = sdk.get_agent_by_name('Autotagger Agent') required = [r for r in agent.requirements if r.required] # Set inputs — filepaths are auto-converted to Dataset IDs agent.requirements['input_dataset'] = '/data/records.csv' agent.requirements['tag_column'] = 'category' # Run the agent (returns True on successful initiation) success = agent.run() # Poll until the job is complete while not agent.retrieve_results(): time.sleep(10) # Access structured output as a Pandas DataFrame df = agent.output_objects[0].get_dataframe() # Pipe into your own pipeline, database, or API df.to_sql('tagged_records', engine, if_exists='replace')
Validating inputs... ✓ OK Uploading dataset... ✓ 14,302 rows Dispatching agent job... ✓ Job #8471 queued Running Autotagger Agent... polling every 10s Job #8471 complete. ✓ 14,302 rows returned Writing to tagged_records...✓ Done
Once you've identified the agent you need, running it is straightforward: provide the required inputs, invoke the agent, and handle the structured output in your code. Agent inputs and outputs are typed — the SDK tells you exactly what each agent needs and what it returns — so there's no guesswork about how to wire things together.
The SDK is designed to fit inside your existing engineering infrastructure — not replace it. Use it to add AI-powered data capabilities to existing pipelines, scheduled jobs, or internal tools without requiring teams to work in the Redbird UI. Outputs land wherever you need them.
from airflow.decorators import dag, task from redbird_agent_sdk import RedbirdAgentSDK import time @dag(schedule_interval='0 6 * * *') def daily_tagging_pipeline(): @task() def run_autotagger(): sdk = RedbirdAgentSDK( user_id=get_secret('REDBIRD_USER_ID'), project_folder_id=get_secret('REDBIRD_PROJECT_ID') ) agent = sdk.get_agent_by_name('Autotagger Agent') agent.requirements['input_dataset'] = fetch_latest_export() agent.run() while not agent.retrieve_results(): time.sleep(10) load_to_warehouse( agent.output_objects[0].get_dataframe() ) run_autotagger() daily_tagging_pipeline()
The Redbird Agent SDK is a Python package. It works in any environment that runs Python — notebooks, scripts, containerized pipelines, or cloud functions.
Your first agent call is a few lines of Python away.