This is a sample agent designed to demonstrate returning complex data as output from an agent.
create the venv:
uv venv
source .venv/bin/activateinstall/update deps:
uv pip compile requirements.in -o requirements.txt
uv pip install -r requirements.txtensure the Flyte Sandbox is running w/
flytectl demo startbuild project and docker image, then push:
uv build
docker buildx build --platform linux/amd64 \
-t localhost:30000/flyteagent:local \
--build-arg PYTHON_VERSION=3.11 \
--build-arg FLYTEKIT_VERSION=1.15.1 \
-f Dockerfile . --load
docker push localhost:30000/flyteagent:localupdate deployment, point to local image:
k set image deployment/flyteagent flyteagent=localhost:30000/flyteagent:local
k patch deployment flyteagent -p '{"spec": {"template": {"spec": {"containers": [{"name": "flyteagent", "imagePullPolicy": "Always"}]}}}}'restart deployment:
k rollout restart deployment flyteagentrun example:
pyflyte run --remote --copy all -p flytesnacks -d development src/foo.py foo_wf