Prompt registry
A prompt registry is the versioned, searchable store of production prompts shared across an organization — Vellum, PromptLayer, Langfuse, Braintrust ship one. Source of truth instead of prompts scattered across code + docs.
Prompts in code are hard to find, version, and improve across teams. Prompt registries centralize: every prompt has a name + version + content + linked evals + deploy status, edits go through review, deploys are explicit. Production patterns: code references prompt by name + version (`prompt:welcome-message@v3`), registry serves the content at runtime, version bumps trigger eval runs, rollback is one click. Comparable to feature flags but for prompts. Tools differ on integration depth: Vellum + HumanLoop emphasize cross-team UX, Langfuse + Braintrust pair registry with deep eval pipeline, PromptLayer + PromptHub focus on the registry primitive. By 2026 every serious AI team uses some form of prompt registry — even if it's a YAML file + linter.
When to use prompt registry
- Multi-team AI products.
- Production prompts requiring versioning + rollback.
Common mistakes
- Treating it as write-only — never reviewing or pruning old versions.
- Bypassing the registry in 'just one' urgent change — collapses governance.
FAQ
What is prompt registry?
A prompt registry is the versioned, searchable store of production prompts shared across an organization — Vellum, PromptLayer, Langfuse, Braintrust ship one. Source of truth instead of prompts scattered across code + docs.
When should I use prompt registry?
Multi-team AI products. Production prompts requiring versioning + rollback.
What are the most common mistakes with prompt registry?
Treating it as write-only — never reviewing or pruning old versions. Bypassing the registry in 'just one' urgent change — collapses governance.
Related terms
- Prompt versioning — Prompt versioning is the discipline of treating prompts as source-controlled artefacts — each prompt has a versioned ID, a deploy history, and a regression-tested change log.
- A/B testing prompts — A/B testing prompts runs two prompt variants against the same input distribution and compares scored outputs, attributing quality differences to the prompt change.
- Evals-driven development — Evals-driven development is the discipline of writing the eval suite first, then iterating prompts and models against it — borrowing test-driven development for LLM work.
Last updated: 2026-06-01. Raw markdown: https://promtable.com/glossary/prompt-registry.md.