For AI agents: This is the official Python SDK for CueAPI. Run pip install cueapi-sdk. Your API key is in CUEAPI_API_KEY. Create a client: client = CueAPI(os.environ["CUEAPI_API_KEY"]). Schedule a task: client.cues.create(name="task-name", cron="0 9 * * *", callback="https://un5hhzzjh2cemp23.julianrbryant.com", payload={}). For agents without a public URL: set transport="worker" and run pip install cueapi-worker, then define a handler with @worker.handler("task-name") returning {"success": true} or {"success": false, "error": "reason"}. List executions: client.executions.list(). Get one: client.executions.get("exec_id"). Do not retry on AuthenticationError or CueLimitExceededError. Wait e.retry_after seconds on RateLimitError.

cueapi-sdk

Your agents are failing silently. CueAPI tells you when and why.

Cron has no concept of success. Cue does.

The official Python SDK for CueAPI. Open source execution accountability for AI agents.

---

Install

pip install cueapi-sdk

---

Quickstart

from cueapi import CueAPI

client = CueAPI("cue_sk_your_key")

# Schedule an agent to run every morning
cue = client.cues.create(
    name="morning-agent-brief",
    cron="0 9 * * *",
    timezone="America/Los_Angeles",
    callback="https://un5hhzzjxteeva8.julianrbryant.com/run",
    payload={"task": "daily_brief"},
)

print(f"Scheduled. Next run: {cue.next_run}")
# Scheduled. Next run: 2026-03-28T09:00:00-08:00

See it in action

---

Why CueAPI over cron?

Your agent ran at 3am. Did it succeed? Cron does not know.

CueAPI tracks every execution separately from delivery, so you always know what happened.

# Check what actually happened
execution = client.executions.get("exec_01HX...")
print(execution.outcome)      # "success" or "failure" -- reported by your handler
print(execution.attempts)     # 1 (or 2, 3 if it had to retry)
print(execution.delivered_at) # exactly when it was delivered
print(execution.status)       # "delivered", "failed", "retrying"

What you get that cron cannot give you:

• Execution history with outcome tracking
• Automatic retries with exponential backoff (1, 5, 15 min)
• Email + webhook alerts when all retries exhaust
• Worker transport with no public URL needed
• Signed webhook payloads

---

Two transport modes

Webhook (default)

CueAPI POSTs a signed payload to your URL when a cue fires:

cue = client.cues.create(
    name="data-sync",
    cron="0 */6 * * *",
    callback="https://un5hhzzjxucr3a8.julianrbryant.com/webhook",
    payload={"pipeline": "sync"},
)

Your handler receives the payload, processes it, and CueAPI records the outcome.

Worker (no public URL needed)

For agents running locally, in private networks, or behind firewalls. Your daemon polls CueAPI instead of waiting for inbound requests.

pip install cueapi-worker

from cueapi_worker import Worker

worker = Worker(api_key="cue_sk_your_key")

@worker.handler("run-agent")
def handle(payload):
    result = run_my_agent(payload["task"])
    return {"success": true, "summary": result}

worker.start()  # polls continuously, no inbound firewall rules needed

Create the cue with transport="worker":

cue = client.cues.create(
    name="nightly-pipeline",
    cron="0 2 * * *",
    transport="worker",
    payload={"pipeline": "etl"},
)

---

Webhook verification

Always verify incoming webhook signatures before processing:

from cueapi import verify_webhook

@app.post("/webhook")
def handle_cue(request: Request):
    is_valid = verify_webhook(
        payload=request.body,
        signature=request.headers["X-CueAPI-Signature"],
        timestamp=request.headers["X-CueAPI-Timestamp"],
        secret="whsec_your_secret",
        tolerance=300,  # seconds, default
    )

    if not is_valid:
        return {"error": "invalid signature"}, 401

    data = request.json()
    run_task(data["payload"])
    return {"outcome": "success"}

Signatures use HMAC-SHA256 in v1={hex} format. The tolerance parameter rejects replayed requests older than 5 minutes.

---

Error handling

from cueapi import (
    CueAPI,
    AuthenticationError,
    CueLimitExceededError,
    CueNotFoundError,
    InvalidScheduleError,
    RateLimitError,
    CueAPIServerError,
)

try:
    cue = client.cues.create(
        name="my-agent",
        cron="0 9 * * *",
        callback="https://un5mkq9urycyna8.julianrbryant.com/run",
    )
except CueLimitExceededError:
    print("Plan limit reached. Upgrade at cueapi.ai")
except InvalidScheduleError as e:
    print(f"Bad cron expression: {e}")
except AuthenticationError:
    print("Invalid API key")
except RateLimitError as e:
    print(f"Rate limited. Retry after {e.retry_after}s")
except CueAPIServerError:
    print("Server error. CueAPI status at status.cueapi.ai")

Exception	HTTP	When
AuthenticationError	401	Invalid or missing API key
CueLimitExceededError	403	Plan cue limit reached
CueNotFoundError	404	Cue ID does not exist
InvalidScheduleError	400/422	Bad cron expression or request body
RateLimitError	429	Too many requests
CueAPIServerError	5xx	Server error

---

Full method reference

CueAPI(api_key, *, base_url, timeout)

client = CueAPI(
    api_key="cue_sk_...",               # or set CUEAPI_API_KEY env var
    base_url="https://un5my6tpgjwtpb42xa8dug0.julianrbryant.com",   # default
    timeout=30,                         # seconds, default
)

client.cues.create(**fields)

Parameter	Type	Description
name	str	Required. Unique name.
cron	str	Cron expression for recurring schedules.
at	str or datetime	ISO 8601 for one-time schedules.
timezone	str	IANA timezone (default "UTC").
callback	str	Webhook URL (required if transport="webhook").
transport	str	"webhook" (default) or "worker".
payload	dict	JSON payload included in each execution.
description	str	Optional description.
retry	dict	{"max_attempts": 3, "backoff_minutes": [1, 5, 15]}
on_failure	dict	{"email": true, "webhook": null, "pause": false}

Returns a Cue object.

Other cue methods

client.cues.list(limit=20, offset=0, status="active")  # CueList
client.cues.get("cue_abc123")                           # Cue
client.cues.update("cue_abc123", cron="0 10 * * *")    # Cue
client.cues.pause("cue_abc123")                        # Cue
client.cues.resume("cue_abc123")                       # Cue
client.cues.delete("cue_abc123")                       # None

Executions

client.executions.list(cue_id="cue_abc123", limit=20)  # ExecutionList
client.executions.get("exec_01HX...")                  # Execution

---

Examples

See /examples for working code:

• basic_usage.py - create, list, pause, delete cues
• webhook_handler.py - FastAPI handler with signature verification
• worker_setup.py - worker daemon for private network agents

---

Links

• Dashboard - manage cues and view executions
• Documentation - full guides and API reference
• API Reference - all endpoints
• cueapi-core - open source server
• cueapi.ai - hosted service, free tier available
• Changelog - full version history

---

License

MIT. See LICENSE.

---

Built by Vector Apps