Temporal
Durable execution platform for scalable applications
About
Temporal is an open-source durable execution platform that runs units of application logic called Workflows so they survive crashes, restarts, and intermittent failures — execution state is stored as an append-only event history and replayed to recover. It provides automatic retries, timeouts, timers, signals, child workflows, and scheduling, and is widely used to orchestrate long-running AI agent runs (e.g. OpenAI Agents SDK and Pydantic AI both ship Temporal integrations). The server is a Go binary (MIT licensed) with official SDKs in Python, Go, Java, TypeScript, .NET, and PHP, and a managed Temporal Cloud option exists for zero-ops deployments.
What it can do
- • Durable execution: workflows survive process/server crashes via event-sourced history and replay
- • Automatic retries, timeouts, timers, signals, queries, and cron-like schedules
- • Child workflows, continue-as-new, and workflow versioning
- • Deterministic workflow code with side-effect-free guarantees; activities handle I/O
- • Namespaces for multi-tenancy; pluggable data-converter codecs for payload encryption
- • Web UI, CLI, metrics, and OpenTelemetry tracing
- • SDKs in Python, Go, Java, TypeScript, .NET, PHP (plus Ruby/Rust/Scala under temporalio org)
- • Agent integrations: OpenAI Agents SDK, Pydantic AI TemporalDurability, LangGraph, Strands, Google ADK
Setup
- Install the CLI (macOS:
brew install temporal; other platforms: download the release binary). - Start a local dev server:
temporal server start-dev— this gives you a working service plus a web UI for inspecting executions. - Install your SDK, e.g. Python:
pip install temporalio. - Write a Workflow (deterministic orchestration code) and Activities (side-effecting units, retried automatically), then run a Worker process that hosts them.
- Start a Workflow execution from a client and watch it in the UI; kill the worker mid-run and it resumes automatically.
- Production: self-host the cluster (PostgreSQL/Cassandra persistence) or use Temporal Cloud.
Prerequisites
Temporal CLI; an SDK toolchain (Python 3.9+, Go, Node 18+, JDK, .NET, or PHP depending on language); Docker for production cluster deployment; PostgreSQL or Cassandra for self-hosted persistence.
Use cases
Works with
Trust score — 9/10
Scored 9/10 (verified tier, metrics-v1). Evidence: 23,639 GitHub stars (log-scaled vs catalog peers); blended with ~38,050,331 monthly package downloads; last push 0d ago; 30 releases/12mo; 1060 open issues. Manual review: mature (Cadence lineage); self-host ops complexity; dev server ships without auth. License: MIT.
★ 23,639 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.7/10
- Maintenance · recency, releases, issue load
- 8.2/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 9.0/10
- · 23,639 GitHub stars (log-scaled vs catalog peers)
- · blended with ~38,050,331 monthly package downloads
- · last push 0d ago; 30 releases/12mo; 1060 open issues
- · manual review: mature (Cadence lineage); self-host ops complexity; dev server ships without auth
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.