kestrel-workflows 0.34.0

Python SDK for Kestrel Workflows

Kestrel SDK

Python SDK for Kestrel — AI Agents for Platform Engineering.

Build, deploy, and manage workflows programmatically with a typed, fluent API.

Installation

pip install kestrel-workflows

Quick Start

from kestrel import KestrelClient
from kestrel.workflows import Workflow, Trigger, Action

client = KestrelClient(api_key="kestrel_sk_...")

wf = (
    Workflow("Pod Crash RCA + Jira")
    .description("Run RCA on pod crash, create Jira ticket")
    .trigger(
        Trigger.k8s_pod_status()
        .reasons("CrashLoopBackOff")
        .namespace("production")
    )
    .cooldown(hours=24)
    .then(Action.kestrel_trigger_rca().label("Run RCA"))
    .then(Action.jira_create_ticket()
        .project("KAN")
        .title("{{incident.title}}")
        .priority("High")
    )
)

created = client.workflows.deploy(wf, activate=True)
print(f"Deployed: {created.id}")

Async Support

from kestrel import AsyncKestrelClient

async with AsyncKestrelClient(api_key="kestrel_sk_...") as client:
    workflows = await client.workflows.list()
    execution = await client.workflows.test(workflows[0].id)
    result = await client.executions.wait(execution.id)
    print(f"Result: {result.status}")

Integrations

Connect, test, and disconnect Kestrel integrations programmatically. API keys need the integrations:read / integrations:manage scopes.

# See every integration and its connection status
for status in client.integrations.list():
    print(status.id, status.connected)

# Inspect credential requirements per integration
for spec in client.integrations.specs():
    print(spec.key, spec.kind, [f.name for f in spec.fields])

# Setup instructions: where to create each credential (mirrors the
# platform UI and `kestrel integrations connect <name> --help`)
print(client.integrations.setup_help("cloudflare"))

# Token integrations take credential kwargs
client.integrations.connect("cloudflare", api_token="...", account_id="...")
client.integrations.connect("pagerduty", api_token="...", webhook_secret="...")

# Follow-up steps after connecting (e.g. webhook setup), if any
print(client.integrations.post_connect_hint("cloudflare"))

# Save a vendor-generated webhook signing secret after connect
# (Vercel, Railway, PlanetScale, Supabase) — keeps the stored API token
client.integrations.set_webhook_secret("vercel", "whsec_...")

# Knowledge sources
client.integrations.connect(
    "confluence",
    base_url="https://acme.atlassian.net",
    api_key="[e-mail]",   # Atlassian account email
    api_token="...",
)

# OAuth integrations return a URL to open in the browser
url = client.integrations.connect("github")

# Verify and clean up
client.integrations.test("cloudflare")
client.integrations.disconnect("cloudflare")

Kubernetes, AWS, and OCI use multi-step flows — connect those with the Kestrel CLI (kestrel integrations connect <name> --help).

Authentication

Create an API key in the Kestrel platform under Workflows > API Keys.

# API key (recommended)
client = KestrelClient(api_key="kestrel_sk_...")

# From CLI login
client = KestrelClient.from_config()

# Async
client = AsyncKestrelClient(api_key="kestrel_sk_...")

Documentation

Full SDK documentation: docs.usekestrel.ai/workflows/sdk

License

Apache 2.0