PluginBench
MCP Server
Active
MIT

Rails AI Context MCP Server

io.github.crisnahine/rails-ai-context

45 MCP tools give AI agents live Rails schema, routes, models, and conventions—eliminating guessing and correction loops.

What is the Rails AI Context MCP server?

The Rails AI Context MCP server exposes 38 tools and 5 resource templates that query your Rails app's schema, models, routes, controllers, views, and conventions on demand. Every result is verified against your actual codebase using Prism AST parsing and tagged with confidence levels, so AI agents work from ground truth instead of training-data guesses.

Rails AI Context stops AI from making confident mistakes about your codebase. Instead of guessing whether a column exists, an association name is correct, or a method is already implemented, AI agents call tools to verify directly against your live app. This eliminates the correction loop: no more catching errors, fixing them, re-prompting, and breaking something adjacent. Works as an MCP server (stdio or HTTP), CLI tool, or static context files.

How to install Rails AI Context

Copy-paste configuration for popular MCP clients.

transport: stdio
Config generated by PluginBench — verify against the source before use.
~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "rails-ai-context": {
      "command": "https://github.com/crisnahine/rails-ai-context/releases/download/v5.24.0/rails-ai-context-mcp.mcpb",
      "args": []
    }
  }
}

Tools & capabilities

Tools this server exposes to the agent.

  • search_code — Trace method definitions, source code, all callers grouped by type, and tests in one call
  • get_schema — Columns with indexed/unique/encrypted/default hints for any table
  • get_model_details — AST-parsed associations, validations, scopes, enums, macros with [VERIFIED] or [INFERRED] tags
  • get_controllers — Actions with inherited filters, render map, and strong params
  • get_routes — Code-ready route helpers with required params
  • get_view — Templates with instance variables, Turbo wiring, Stimulus refs, and partial locals
  • get_stimulus — HTML data-attributes, targets, values, and actions for Stimulus controllers
  • get_callbacks — Callbacks in Rails execution order with source code
  • get_concern — Concern methods, source, and which models include it
  • analyze_feature — Full-stack analysis: models, controllers, routes, services, jobs, views, and tests
  • get_context — Composite context: schema, model, controller, routes, and views in one call
  • get_test_info — Fixtures, relationships, and test templates matching your patterns
  • generate_test — Test scaffolding that matches your project's patterns
  • validate — Syntax, semantic, and Brakeman security checks in one call
  • security_scan — Brakeman static analysis for SQL injection, XSS, mass assignment
  • performance_check — N+1 risks, missing indexes, counter_cache, and eager load candidates
  • get_conventions — Auth checks, flash messages, create action template, test patterns
  • get_config — Database, auth framework, assets, cache, queue, Action Cable config
  • get_gems — Notable gems with versions, categories, and config file locations
  • get_env — Environment variables and credentials keys (not values)

Use cases

  • Verify schema before writing migrations—catch duplicate columns before they're created
  • Resolve actual association names and method signatures from models instead of guessing
  • Generate tests that match your project's patterns (fixtures vs. factories, test structure)
  • Trace where a method is defined and used across the entire codebase in one call
  • Review code changes with full context about inherited filters, strong params, and conventions

Rails AI Context MCP server FAQ

What is the Rails AI Context MCP server?

It's a Model Context Protocol server that exposes 38 tools and 5 resource templates to query your Rails app's schema, models, routes, controllers, views, and conventions. Every result is verified against your actual codebase using Prism AST parsing, so AI agents work from ground truth instead of guesses.

Is it free?

Yes. Rails AI Context is open-source under the MIT license. The gem is free to install and use.

How do I install it in Cursor or Claude?

Install the gem (`gem install rails-ai-context`), run `rails-ai-context init` for interactive setup, then `rails-ai-context serve` to start the MCP server. Cursor and Claude will auto-detect the server config file on project open. Alternatively, add to your Gemfile and run `rails generate rails_ai_context:install`.

Does it require authentication or API keys?

No. It's a local tool that introspects your Rails app directly. No external services, no API keys, no authentication required.

What Rails versions does it support?

Rails 7.1, 7.2, and 8.0 with Ruby 3.2, 3.3, or 3.4.

Can I use it without an MCP server—just the CLI?

Yes. All 38 tools are available via CLI: `rails 'ai:tool[search_code]' pattern=your_method match_type=trace`. No server needed.

README (reference)

Source of truth, from the repository.

<div align="center">

rails-ai-context

Your AI is guessing your Rails app. Every guess costs you time.

<a href="https://claude.ai/claude-code"><img src="https://img.shields.io/badge/Claude_Code-ee8b4a?style=for-the-badge&logo=anthropic&logoColor=white" alt="Claude Code"></a> <a href="https://cursor.com"><img src="https://img.shields.io/badge/Cursor-000000?style=for-the-badge&logo=cursor&logoColor=white" alt="Cursor"></a> <a href="https://github.com/features/copilot"><img src="https://img.shields.io/badge/GitHub_Copilot-000000?style=for-the-badge&logo=githubcopilot&logoColor=white" alt="GitHub Copilot"></a> <a href="https://opencode.ai"><img src="https://img.shields.io/badge/OpenCode-4285F4?style=for-the-badge&logoColor=white" alt="OpenCode"></a> <a href="https://codex.openai.com"><img src="https://img.shields.io/badge/Codex_CLI-412991?style=for-the-badge&logo=openai&logoColor=white" alt="Codex CLI"></a> <a href="docs/CLI.md"><img src="https://img.shields.io/badge/Any_Terminal-4EAA25?style=for-the-badge&logo=gnubash&logoColor=white" alt="Any Terminal"></a>

Gem Version Downloads CI MCP Registry <br> Ruby Rails Tests License: MIT

</div>

The problem

You've seen it. Your AI:

  • Writes a migration for a column that already exists - didn't check the schema
  • Creates a method that duplicates one in a concern - didn't know it was there
  • Uses the wrong association name - user.posts when it's user.articles
  • Generates tests that don't match your patterns - factories when you use fixtures, or the reverse
  • Adds a gem you already have - or calls an API from one you don't
  • Misses before_action filters from parent controllers - then wonders why auth fails
  • Invents a method that isn't in your codebase - then you spend 10 minutes finding out

You catch it. You fix it. You re-prompt. It breaks something else.

The real cost of AI coding isn't the tokens - it's the correction loop. Every guess is a round-trip: you catch it, you fix it, you re-prompt, and something adjacent breaks. This gem kills the guessing at its source.

<br>

Two commands. Problem gone.

gem "rails-ai-context", group: :development
rails generate rails_ai_context:install

Or standalone - no Gemfile needed

gem install rails-ai-context
cd your-rails-app
rails-ai-context init     # interactive setup
rails-ai-context serve    # start MCP server
<div align="center">

Install demo

</div>

Now your AI doesn't guess - it asks your app directly. 38 tools and 5 resource templates that query your schema, models, routes, controllers, views, and conventions on demand. Model introspection uses Prism AST parsing - every result carries a [VERIFIED] or [INFERRED] confidence tag so AI knows what's ground truth and what needs runtime checking.

<br>

See the difference

<div align="center">

Trace demo

</div>

One call returns: definition + source code + every caller grouped by type + tests. Replaces 4-5 sequential file reads.

<br>

What stops being wrong

Real scenarios where AI goes sideways - and what it does instead with ground truth:

You ask AI to...Without - AI guessesWith - AI verifies first
Add a subscription_tier column to usersWrites the migration, duplicates an existing columnReads live schema, spots subscription_status already exists, asks before migrating
Call user.posts in a controllerUses the guess; runtime NoMethodErrorResolves the actual association (user.articles) from the model
Write tests for a new modelScaffolds with FactoryBotDetects your fixture-based suite and matches it
Fix a failing create actionMisses inherited before_action :authenticate_user!Returns parent-controller filters inline with the action source
Build a dashboard pageInvents Tailwind classes from memoryReturns your actual button/card/alert patterns, copy-paste ready
Trace where publishable? is usedReads 6 files sequentially, still misses callersSingle call: definition + source + every caller + tests
<details> <summary><strong>Verify it on your own app</strong></summary> <br>

Run these before and after installing to see what changes in your codebase:

# Schema: does AI know what columns exist?
rails 'ai:tool[schema]' table=users

# Trace: find every caller of a method across the codebase
rails 'ai:tool[search_code]' pattern=your_method match_type=trace

# Model: associations, scopes, callbacks, concerns - all resolved
rails 'ai:tool[model_details]' model=User

# Controllers: action source + inherited filters + strong params in one shot
rails 'ai:tool[controllers]' controller=UsersController action=create

Compare what AI outputs with and without these tools wired in. The difference is measured in corrections avoided, not bytes saved.

</details> <br>

Three ways to use it

<table> <tr> <td width="33%">

MCP Server (stdio)

AI calls tools directly via the protocol. Each AI tool gets its own config file - auto-detected on project open.

rails ai:serve
→ rails_search_code(pattern: "publishable?", match_type: "trace")
→ rails_get_schema(table: "users")
→ rails_analyze_feature(feature: "billing")
</td> <td width="33%">

MCP Server (HTTP)

Mount inside your Rails app - inherits routing, auth, and middleware.

# config/routes.rb
mount RailsAiContext::Engine, at: "/mcp"

Native Rails controller transport. No separate process needed.

</td> <td width="33%">

CLI

Same 38 tools, no server needed. Works in any terminal, any AI tool.

rails 'ai:tool[search_code]' pattern="publishable?" match_type=trace
rails 'ai:tool[schema]' table=users
rails 'ai:tool[analyze_feature]' feature=billing
</td> </tr> </table>

Full Guide → - every command, every parameter, every configuration option.

<br>

Real-world examples

<details> <summary><strong>"Add a subscription field to users"</strong></summary> <br>
rails 'ai:tool[schema]' table=users
## Table: users
| Column              | Type    | Null | Default |
|---------------------|---------|------|---------|
| email               | string  | NO   | [unique] |
| subscription_status | string  | yes  | "free"   |
| created_at          | datetime| NO   |          |

AI sees subscription_status already exists. Checks the model, then generates a correct migration - first attempt.

</details> <details> <summary><strong>"Fix the broken post creation flow"</strong></summary> <br>
rails 'ai:tool[controllers]' controller=PostsController action=create
# PostsController#create

Filters: before_action :authenticate_user!, before_action :set_post (only: show, edit)
Strong params: post_params → title, body, published_at
Renders: redirect_to @post | render :new

AI sees the inherited authenticate_user! filter, the actual strong params, and the render paths. No guessing.

</details> <details> <summary><strong>"Build a new dashboard view"</strong></summary> <br>
# Check existing view patterns
rails 'ai:tool[view]' controller=dashboard
# → templates with ivars, Turbo frames, Stimulus controllers, partial locals

# See existing components + usage examples
rails 'ai:tool[component_catalog]' detail=standard
# → ViewComponent/Phlex props, slots, previews, sidecar assets

# Get Stimulus data-attributes
rails 'ai:tool[stimulus]' controller=chart
# → correct HTML with dashes (not underscores) + reverse view lookup
</details> <br>

38 Tools

Every tool is read-only and returns data verified against your actual app - not guesses, not training data.

<details open> <summary><strong>Search & Trace</strong></summary>
ToolWhat it does
search_codeTrace: definition + source + callers + tests. Also: definition, call, class filters
get_edit_contextMethod-aware code extraction with class context
</details> <details open> <summary><strong>Understand</strong></summary>
ToolWhat it does
analyze_featureFull-stack: models + controllers + routes + services + jobs + views + tests
get_contextComposite: schema + model + controller + routes + views in one call
onboardNarrative app walkthrough (quick/standard/full)
</details> <details open> <summary><strong>Schema & Models</strong></summary>
ToolWhat it does
get_schemaColumns with indexed/unique/encrypted/default hints
get_model_detailsAST-parsed associations, validations, scopes, enums, macros - each result tagged [VERIFIED] or [INFERRED]
get_callbacksCallbacks in Rails execution order with source
get_concernConcern methods + source + which models include it
</details> <details open> <summary><strong>Controllers & Routes</strong></summary>
ToolWhat it does
get_controllersActions + inherited filters + render map + strong params
get_routesCode-ready helpers (post_path(@record)) + required params
</details> <details open> <summary><strong>Views & Frontend</strong></summary>
ToolWhat it does
get_viewTemplates with ivars, Turbo wiring, Stimulus refs, partial locals
get_stimulusHTML data-attributes (dashes!) + targets + values + actions
get_partial_interfaceWhat locals to pass + what methods are called on them
get_turbo_mapBroadcast → subscription wiring + mismatch warnings
get_frontend_stackReact/Vue/Svelte/Angular, Hotwire, TypeScript, package manager
</details> <details open> <summary><strong>Testing & Quality</strong></summary>
ToolWhat it does
get_test_infoFixtures + relationships + test template matching your patterns
generate_testTest scaffolding matching your project's patterns
validateSyntax + semantic + Brakeman security in one call
security_scanBrakeman static analysis - SQL injection, XSS, mass assignment
performance_checkN+1 risks, missing indexes, counter_cache, eager load candidates
</details> <details open> <summary><strong>App Config & Services</strong></summary>
ToolWhat it does
get_conventionsAuth checks, flash messages, create action template, test patterns
get_configDatabase, auth framework, assets, cache, queue, Action Cable
get_gemsNotable gems with versions, categories, config file locations
get_envEnvironment variables + credentials keys (not values)
get_helper_methodsApp + framework helpers with view cross-references
get_service_patternInterface, dependencies, side effects, callers
get_job_patternQueue, retries, guard clauses, broadcasts, schedules
get_component_catalogViewComponent/Phlex: props, slots, previews, sidecar assets
</details> <details open> <summary><strong>Data & Debugging</strong></summary>
ToolWhat it does
dependency_graphModel/service dependency graph in Mermaid or text format
migration_advisorMigration code generation with reversibility + affected models
search_docsBundled topic index with weighted keyword search
querySafe read-only SQL with timeout, row limit, column redaction
read_logsReverse file tail with level filtering and sensitive data redaction
diagnoseOne-call error diagnosis with classification + context + git + logs
review_changesPR/commit review with per-file context + warnings
runtime_infoLive DB pool, table sizes, pending migrations, cache stats, queue depth
session_contextSession-aware context tracking across tool calls
</details>

All 38 tools with parameters →  |  Real-world recipes →

<br>

Live Resources (VFS)

AI clients can also read structured data through resource templates - rails-ai-context:// URIs that introspect fresh on every request. Zero stale data.

Resource TemplateWhat it returns
rails-ai-context://controllers/{name}Actions, inherited filters, strong params
rails-ai-context://controllers/{name}/{action}Action source code with applicable filters
rails-ai-context://views/{path}View template content (path traversal protected)
rails-ai-context://routes/{controller}Live route map filtered by controller
rails://models/{name}Per-model details: associations, validations, schema

Plus 9 static resources (schema, routes, conventions, gems, controllers, config, tests, migrations, engines) that AI clients read directly.

<br>

Anti-Hallucination Protocol

Every generated context file ships with 6 rules that force AI verification before writing code. The protocol targets the exact cognitive failures that produce confident-wrong code: statistical priors overriding observed facts, pattern completion beating verification, stale context lies.

<details> <summary><strong>The 6 rules (shown to AI in every CLAUDE.md / .cursor/rules / .cursorrules / .github/instructions)</strong></summary> <br>
  1. Verify before you write. Never reference a column, association, route, helper, method, class, partial, or gem not verified in THIS project via a tool call in THIS turn. Never invent names that "sound right."
  2. Mark every assumption. If proceeding without verification, prefix with [ASSUMPTION]. Silent assumptions forbidden. "I'd need to check X first" is a preferred answer.
  3. Training data describes average Rails. This app isn't average. When something feels "obviously" standard Rails, query anyway. Check rails_get_conventions + rails_get_gems BEFORE scaffolding.
  4. Check the inheritance chain before every edit. Inherited before_action filters, concerns, includes, STI parents. Inheritance is never flat.
  5. Empty tool output is information, not permission. "0 callers found" signals investigation, not license to proceed on guesses.
  6. Stale context lies. Re-query after writes. Earlier tool output may be wrong after edits.

Enabled by default. Disable with config.anti_hallucination_rules = false if you prefer your own rules.

</details> <br>

How it works

graph TD
    A["Your Rails App\nmodels + schema + routes + controllers + views + jobs"] -->|"39 introspectors"| B

    B["rails-ai-context\nPrism AST parsing · Cached · Confidence-tagged\nVFS: rails-ai-context:// URIs introspected fresh"]

    B --> C["MCP Server\nstdio / HTTP\n38 tools · 5 templates"]
    B --> D["CLI Tools\nRake / Thor\nSame 38 tools"]
    B --> E["Static Files\nCLAUDE.md · .cursor/rules/ · .cursorrules\n.github/instructions/"]

    style A fill:#4a9eff,stroke:#2d7ad4,color:#fff
    style B fill:#2d2d2d,stroke:#555,color:#fff
    style C fill:#0984e3,stroke:#0770c2,color:#fff
    style D fill:#00cec9,stroke:#00b5b0,color:#fff
    style E fill:#a29bfe,stroke:#8c83f0,color:#fff
<br>

Install

Option A - In Gemfile:

gem "rails-ai-context", group: :development
rails generate rails_ai_context:install

Option B - Standalone (no Gemfile entry needed):

gem install rails-ai-context
cd your-rails-app
rails-ai-context init

Both paths ask which AI tools you use (Claude Code, Cursor, GitHub Copilot, OpenCode, Codex CLI) and whether you want MCP or CLI mode. Each tool gets its own MCP config file - auto-detected on project open.

<br>

Commands

In-GemfileStandaloneWhat it does
rails ai:contextrails-ai-context contextGenerate context files
rails 'ai:tool[NAME]'rails-ai-context tool NAMERun any of the 38 tools
rails ai:toolrails-ai-context tool --listList all available tools
rails ai:serverails-ai-context serveStart MCP server (stdio)
rails ai:doctorrails-ai-context doctorDiagnostics + AI readiness score
rails ai:watchrails-ai-context watchAuto-regenerate on file changes
<br>

Documentation

Quickstart5-minute getting started
Tools ReferenceAll 38 tools with every parameter
RecipesReal-world workflows and examples
Custom ToolsBuild and test your own MCP tools
Configuration40+ config options with defaults
AI Tool SetupClaude, Cursor, Copilot, OpenCode, Codex
ArchitectureSystem design and internals
IntrospectorsAll 39 introspectors and AST engine
Security4-layer SQL safety and file blocking
CLI ReferenceCommands and argument syntax
StandaloneUse without Gemfile entry
TroubleshootingCommon issues and fixes
FAQFrequently asked questions
<br>

Build your own tools

Register custom MCP tools alongside the 38 built-in ones:

# app/mcp_tools/rails_get_business_metrics.rb
class RailsGetBusinessMetrics < MCP::Tool
  tool_name "rails_get_business_metrics"
  description "Key business metrics for this app"

  def call(period: "week")
    MCP::Tool::Response.new([{ type: "text", text: "Users this #{period}: #{User.recent.count}" }])
  end
end

# config/initializers/rails_ai_context.rb
config.custom_tools = [RailsGetBusinessMetrics]

Test with the built-in TestHelper (works with RSpec and Minitest):

include RailsAiContext::TestHelper

response = execute_tool("business_metrics", period: "month")
assert_tool_response_includes(response, "Users")

Custom Tools Guide →

<br>

Configuration

# config/initializers/rails_ai_context.rb
if defined?(RailsAiContext)
  RailsAiContext.configure do |config|
    config.ai_tools   = %i[claude cursor] # Which AI tools to generate for
    config.tool_mode  = :mcp              # :mcp (default) or :cli
    config.preset     = :full             # :full (39 introspectors) or :standard (17)
  end
end

All 40+ configuration options →

<br>

Observability

Every MCP tool call fires an ActiveSupport::Notifications event:

ActiveSupport::Notifications.subscribe("rails_ai_context.tools.call") do |event|
  Rails.logger.info "[MCP] #{event.payload[:method]} - #{event.duration}ms"
end
<br>

Requirements

  • Ruby >= 3.2    Rails >= 7.1
  • Optional: brakeman for security scanning, listen for watch mode, ripgrep for fast search
<br>
<div align="center">

About

Built by a Rails developer with 10+ years of production experience.<br> 2154 tests + 100-example e2e harness. 38 tools. 5 resource templates. 39 introspectors. Standalone or in-Gemfile.<br> MIT licensed. Contributions welcome.

<br>

If this gem saves you time, consider sponsoring the project.

<br>

MIT License

</div>

Related MCP servers

Read-only GitHub bounty, agent harness, Actions failure, flake, and MCP tool-drift decisions.

0
TypeScript
MIT
View repository →
LILiteScope logo

LiteScope

Active

Operations toolchain for SQLite, Cloudflare D1 & Turso: inspect, diff, migrate, monitor & repair.

1
Go
AGPL-3.0
View repository →

MCP server for Croit Ceph cluster management with dynamic OpenAPI tool generation

1
Python
View repository →

MCP server for generating advanced mock 2D PNG assets for game prototypes.

3
TypeScript
MIT
View repository →

Argentina CUIT/CUIL validation and AFIP fiscal QR generation. Pure, no-auth MCP tools.

Calibrated probabilistic foresight for AI agents, powered by live prediction-market signal.