Inspect resources with Python

Each method’s parameters, defaults, and return types are in the client.benchmarks, client.agents, and client.services reference.

Benchmarks and tasks

Use client.benchmarks for detailed, paginated run views:

from valkyrie.sdk import FetchTasksRequest, TaskStatus

details = await client.benchmarks.fetch(run.benchmark_id)
statuses = await client.benchmarks.statuses([run.benchmark_id])
tasks = await client.benchmarks.tasks(
    run.benchmark_id,
    FetchTasksRequest(status=[TaskStatus.ERROR], limit=25),
)
task = await client.benchmarks.task(run.benchmark_id, tasks.tasks[0].task_id)
artifacts = await client.benchmarks.artifacts(run.benchmark_id, task.task_id)

Artifact methods return temporary URLs generated by Tracker. The SDK does not expose internal S3 keys or clients. Tracker task-detail routes do not support task IDs containing /; the SDK rejects those IDs before sending a request.

Agents

List uploaded agents or request a temporary download URL:

agents = await client.agents.list()
download = await client.agents.download_url(agents.agents[0].name)

Local agent packaging and S3 upload or download remain CLI workflows.

Benchmark services

Discover services, check their health, or fetch dataset task IDs:

services = await client.services.list()
task_ids = await client.services.task_ids("swebench", dataset="default")

Get started

Agents

Runs

Benchmarks

Integrations

Python SDK

CLI reference

Python SDK reference

Self-hosting

Inspect resources with Python

Benchmarks and tasks

Agents

Benchmark services

​Benchmarks and tasks

​Agents

​Benchmark services

Benchmarks and tasks

Agents

Benchmark services