Skip to main content
skill CX Codex

dbos-python — Claude Code skill: durable DBOS workflows

A skill that teaches your coding agent DBOS Python 3.x: workflows, steps, queues, datasource transactions and the 2.x upgrade.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add dbos-inc/agent-skills --skill dbos-python -a claude-code
Open on GitHub

What dbos-python does

This skill feeds the agent the rules of DBOS Python 3.x while it writes or edits your code. It ships as reference files split by category: lifecycle, workflow, step, queue, communication, patterns, testing, client and advanced. The agent opens the file it needs instead of guessing at the API.

The rules carry priorities. The critical ones cover launch: a DBOS app must configure and launch DBOS inside its main function, and workflows have to stay deterministic, so anything non-deterministic belongs in a step. A separate list covers what 3.0 removed and the agent should never generate: @DBOS.transaction, DBOS.sql_session, application_database_url, Queue(...), @DBOS.scheduled, and the DBOS(fastapi=...) / DBOS(flask=...) constructors.

Two ordering traps get their own mention. Register queues and schedules after DBOS.launch(), create datasources before it. In async def code you need the _async variants — synchronous calls like DBOS.sleep, DBOS.recv, DBOS.send, DBOS.set_event, DBOS.get_event and DBOS.register_queue raise RuntimeError when an event loop is running. Connecting to DBOS Conductor needs pip install "dbos[enterprise]", so check that dependency before you wire it up.

What it helps with

  • split a workflow into steps around external calls
  • cap concurrency with a registered queue
  • migrate DBOS Python 2.x code to 3.x
  • call a workflow from an outside app via DBOSClient
  • fix a synchronous DBOS method failing inside async code

Example prompts for the agent

  • add DBOS to this project and sketch a workflow with steps
  • register a queue with a concurrency limit after DBOS.launch()
  • help me move this file from DBOS 2.x to 3.x

How to install dbos-python

1Skills CLI (Vercel)

npx skills add dbos-inc/agent-skills --skill dbos-python -a claude-code

2Manually: put the files in place

~/.claude/skills/dbos-python/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/dbos-python/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about dbos-python

How do I install dbos-python in Claude?

Run npx skills add dbos-inc/agent-skills --skill dbos-python -a claude-code. The skill lands in ~/.claude/skills/dbos-python/SKILL.md.

What does dbos-python do?

It gives the agent DBOS Python 3.x rules: configuring and launching an app, writing workflows and steps, using queues, schedules and DBOSClient. One reference file covers the 2.x upgrade.

Which DBOS methods break in async code?

Synchronous DBOS.sleep, DBOS.recv, DBOS.send, DBOS.set_event, DBOS.get_event and DBOS.register_queue raise RuntimeError while an event loop runs. Use the _async variants inside async def.

Source text

SKILL.md · MIT
---
name: dbos-python
description: DBOS Python SDK (3.x) for building reliable, fault-tolerant applications with durable workflows. Use this skill when writing Python code with DBOS, creating workflows and steps, using queues, datasource transactions, or schedules, using DBOSClient from external applications, upgrading DBOS Python 2.x code to 3.x, or building applications that need to be resilient to failures.
license: MIT
metadata:
  author: dbos
  version: "2.1.0"
  organization: DBOS
  date: October 2026
  abstract: Comprehensive guide for building fault-tolerant Python applications with DBOS. Covers workflows, steps, queues, communication patterns, and best practices for durable execution.
---

# DBOS Python Best Practices

Guide for building reliable, fault-tolerant Python applications with DBOS durable workflows. Targets DBOS Python 3.x.

## When to Apply

Reference these guidelines when:
- Adding DBOS to existing Python code
- Creating workflows and steps
- Using queues for concurrency control
- Implementing workflow communication (events, messages, streams)
- Configuring and launching DBOS applications
- Using DBOSClient from external applications
- Testing DBOS applications
- Upgrading DBOS Python 2.x code to 3.x

## Rule Categories by Priority

| Priority | Category | Impact | Prefix |
|----------|----------|--------|--------|
| 1 | Lifecycle | CRITICAL | `lifecycle-` |
| 2 | Workflow | CRITICAL | `workflow-` |
| 3 | Step | HIGH | `step-` |
| 4 | Queue | HIGH | `queue-` |
| 5 | Communication | MEDIUM | `comm-` |
| 6 | Pattern | MEDIUM | `pattern-` |
| 7 | Testing | LOW-MEDIUM | `test-` |
| 8 | Client | MEDIUM | `client-` |
| 9 | Advanced | LOW | `advanced-` |

## Critical Rules

### DBOS Configuration and Launch

A DBOS application MUST configure and launch DBOS inside its main function:

```python
import os
from dbos import DBOS, DBOSConfig

@DBOS.workflow()
def my_workflow():
    pass

if __name__ == "__main__":
    config: DBOSConfig = {
        "name": "my-app",
        "application_version": "0.1.0",
        "system_database_url": os.environ.get("DBOS_SYSTEM_DATABASE_URL"),
    }
    DBOS(config=config)
    DBOS.launch()
```

When creating a new application, set `application_version` to `"0.1.0"`. If omitted, DBOS derives an opaque hash from workflow source code. When editing an existing application, leave its configured version alone — changing it is a deployment decision (see `references/advanced-versioning.md`).

### Workflow and Step Structure

Workflows are comprised of steps. Any function performing complex operations or accessing external services must be a step:

```python
@DBOS.step()
def call_external_api():
    return requests.get("https://api.example.com").json()

@DBOS.workflow()
def my_workflow():
    result = call_external_api()
    return result
```

### Key Constraints

- Do NOT call `DBOS.start_workflow` or `DBOS.recv` from a step
- Do NOT use threads to start workflows - use `DBOS.start_workflow` or queues
- Workflows MUST be deterministic - non-deterministic operations go in steps
- Do NOT create/update global variables from workflows or steps
- In `async def` code, use the `_async` variants of DBOS methods (`await DBOS.start_workflow_async(...)`, `send_async`, `recv_async`, `sleep_async`, `register_queue_async`, ...): many synchronous DBOS methods (such as `DBOS.sleep`, `DBOS.recv`, `DBOS.send`, `DBOS.set_event`, `DBOS.get_event`, and `DBOS.register_queue`) raise `RuntimeError` when called while an event loop is running
- Register queues and create schedules AFTER `DBOS.launch()`; create datasources BEFORE it
- Connecting to DBOS Conductor (`conductor_key`, or deploying to DBOS Cloud) requires `pip install "dbos[enterprise]"`

### Removed in DBOS 3.0 (never generate these)

`@DBOS.transaction` / `DBOS.sql_session` / `application_database_url` (use datasources), `Queue(...)` (use `DBOS.register_queue`), `partition_queue` / `priority_enabled`, `@DBOS.scheduled` (use `DBOS.apply_schedules`), and `DBOS(fastapi=...)` / `DBOS(flask=...)`. To migrate existing 2.x code, see `references/advanced-upgrading-v3.md`.

## How to Use

Read individual rule files for detailed explanations and examples:

```
references/lifecycle-config.md
references/workflow-determinism.md
references/queue-concurrency.md
references/advanced-upgrading-v3.md
references/advanced-shared-database.md
```

If multiple applications share one system database, see `references/advanced-shared-database.md`.

## References

- https://docs.dbos.dev/
- https://github.com/dbos-inc/dbos-transact-py

© dbos · MIT · dbos-inc/agent-skills

Similar skills

Everything for Backend and APIs →

From the same set2

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor