Contentrain AI is a repo-native content governance tool for AI agents, with the goal of extracting, governing, and delivering UI text, documentation, and structured content from a codebase. It helps teams turn hardcoded UI text, documentation, and structured content into a governable, reviewable content layer. Agents can extract, create, translate, and update content, while Contentrain enforces schemas, Git review, and portable file output.
Purpose
Contentrain is mainly used in the following scenarios:
- Rescuing hardcoded strings from existing applications and turning them into structured, translatable content;
- Starting new projects with a true content layer, instead of scattering copy across components;
- Serving the same content from plain JSON and Markdown in Git to web, documentation, mobile, and backend systems.
It is explicitly not an AI writing application, not a database-first or dashboard-first CMS replacement for all teams, and not a proprietary content runtime that locks an application to a single platform.
Main capabilities
- Normalize workflow: Scans a codebase for hardcoded strings, extracts content, creates i18n-ready content, and patches source files, all through a reviewable Git branch.
- MCP engine: Provides 27 tools (22 core tools + 5 media tools), supports stdio or HTTP transport, and can be used with Claude Code, Cursor, Windsurf, or any MCP client.
- Provider-agnostic engine: The same tool interface can run on a local worktree, GitHub, or GitLab (including self-hosted), without modifying tool code. HTTP transport can be used for remote driving such as Studio.
- Canonical serialization: Key sorting, deterministic output, clean git diffs, conflict-free parallel editing.
- Agent rules and skills: Behavior policies and step-by-step workflows are published as npm packages.
- Serve UI: A local web dashboard for browsing models, content, validation, and normalize status.
- Framework-agnostic: MCP is not aware of specific frameworks; agents and skills handle stack-specific logic.
- Content types and fields: Supports 4 content types and 27 field types (string, email, url, image, relation, array, object, markdown, etc.), with built-in validation.
- Typed SDK: content is pure JSON and Markdown, directly readable in any language; TypeScript projects can generate a typed SDK.
- Multi-stack support: works with 20+ tech stacks including Nuxt, Next.js, Astro, SvelteKit, Vue, React, Node, Go, Python, Swift, Flutter, and more.
How to Use
The repository provides 16 Agent Skills, i.e. reusable workflow processes, that any AI coding agent can load on demand. Install all skills into the agent, or install specific skills. The skills can be used with 40+ agents including Claude Code, Cursor, Windsurf, GitHub Copilot, OpenAI Codex, Gemini CLI, and more. For the full skill catalog and agent guide, see AGENTS.md.
Quick-start references include: a 2-minute demo, a getting-started guide (installation, connecting an agent, running your first workflow), the Normalize guide (the main hardcoded-string rescue process), the ecosystem map, Contentrain Studio (open-source core team operations for Git-native structured content, self-hostable or available as hosted Starter/Pro/Enterprise plans), and full documentation.
The repository also provides production-ready starter templates, including Contentrain content models, generated SDK clients, and framework-specific patterns, each of which is a GitHub template repository. Among them, templates/astro-starter is located in this repository and is a starting point for WordPress migration, covering posts, pages, archives and permalinks, SEO, RSS, sitemap, search, Studio forms and comments, Gutenberg block styles, and passes its own gates (astro check, knip, build, built-site checks, Lighthouse) with no content.
Applicable Scenarios
Suitable for development teams that need to centralize scattered copy, need multilingual content, need a Git review process, and want content to be reused across frameworks as plain files.
Limitations
The materials do not provide specific pricing, platform limitations, or performance metrics. Contentrain Studio offers an open-source core and hosted plans (Starter/Pro/Enterprise), but specific prices are not yet provided.
