encord-agents 0.1.0

A framework for easily producing custom agens for your Encord Workflows.

Documentation | Try it Now | Blog | Join our Community

Easily build agents for the Encord echo system. With just few lines of code, you can take automation to the next level.

Key features:

1. ⚡Easy: Multiple template agents to be adapted and hosted via GCP, own infra, or cloud.
2. ⏩ Convenient: The library conveniently loads data via the Encord SDK upon request.
3. 👨‍💻 Focus: With essential resources readily available, you can focus on what matters. Create agents with pre-existing (or custom) dependencies for loading labels and data.
4. 🤏 Slim: the library is slim at it's core and should not conflict with the dependencies of most projects.

💡 For the full documentation and end-to-end examples, please see here.

Here are some use-cases:

Here's how to build an Agent:

from uuid import UUID
from encord.objects.ontology_labels_impl import LabelRowV2
from encord_agents.tasks import Runner

runner = Runner(project_hash="<your_project_hash>")


@runner.stage(UUID("<your_agent_stage_uuid>"))
def by_file_name(lr: LabelRowV2) -> UUID | None:
    # Assuming the data_title is of the format "%d.jpg"
    # and in the range [0; 100]
    priority = int(lr.data_title.split(".")[0]) / 100
    lr.set_priority(priority=priority)
    return UUID("<your_pathway_uuid>")


if __name__ == "__main__":
    runner.run()

You can also inject dependencies:

from typing_extensions import Annotated

from encord.objects import LabelRowV2
from encord_agents.tasks import Runner, Depends

runner = Runner(project_hash="<your_project_hash>")

def my_custom_dependency(label_row: LabelRowV2) -> dict:
    # e.g., look up additional data in own db
    return db.query("whatever")

@runner.stage(stage="<my_stage_name>")
def by_custom_data(
    custom_data: Annotated[dict, Depends(my_custom_dependency)]
) -> str:
    # `custom_data` automatically injected here.
    # ... do your thing
    # then, return name of task pathway.


if __name__ == "__main__":
    runner.run()

Please visit our 📖 Documentation for a complete reference to how to use the agents library.