rails-ai-context 0.10.1 → 0.10.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 41d6a7f273bd4630ef25d299a4379609336714a021d77bcbb5128235a62bd830
-  data.tar.gz: 64b3c29bbf29126d6f8aaf76e1f063140ad7ea3684e15ca392a2275d2fe35110
+  metadata.gz: 68c0816e248242dd0b6b6c122f8a5538295823a4da7386746aa3d3ea8c3a8ee0
+  data.tar.gz: 178f97d3d8557621dbe0cc163e75e83afed3156ca1ea87d185e29ca380567580
 SHA512:
-  metadata.gz: 90003df73a6d12bec5e205343ffc4615a6f59061f32bdba71f344dfcacf046de339836b871d1e6e85353bff3951673a5fa2f2c2b0bd3b950c9b0d62872760bc2
-  data.tar.gz: eaad1515c31ee3d6d3cb61a402991d92aa6107770011d70338a096d8dc23a809423e77ba3194d7813cbb3df2ea3cadf772ebdf4f7b939afe0c82940471b7b7a3
+  metadata.gz: 953fcec85a05ec05862157ae70a8715f0daad9d1b4f1fba075c8f1ef16a13de90eaf81e3906efd5790fb4ed32106bb5838668ff0d41533a569a7e6443f63b141
+  data.tar.gz: eade81b4fa9597317981d8f17a6fccdc6d9872adae3b543dd18cf9449d6746072584f754ded5031c2376b33d30aded7164c03f7253513e1c2f2c7b9c4c35d225

data/README.md

@@ -21,22 +21,49 @@ The AI doesn't know your schema, your Devise setup, your Sidekiq jobs, or that `
 
 ---
 
-## Proof: 35% Token Savings (Real Benchmark)
+## Proof: 37% Token Savings (Real Benchmark)
 
-We ran the same feature task — *"Add status and date range filters to the Cooks index page"* — across 4 scenarios in parallel on a real Rails app:
+Same task — *"Add status and date range filters to the Cooks index page"* — 4 scenarios in parallel, same Rails app:
 
-| Scenario | MCP Tools | CLAUDE.md | Tokens Used | Savings |
-|----------|-----------|-----------|-------------|---------|
-| **MCP + CLAUDE.md** | Yes | Yes | **25,884** | **35% saved** |
-| MCP only | Yes | No | 27,822 | 31% saved |
-| CLAUDE.md only | No | Yes | 32,699 | 19% saved |
-| Zero (nothing) | No | No | 40,129 | baseline |
+| Setup | Tokens | Saved | What it knows |
+|-------|--------|-------|---------------|
+| **rails-ai-context (full)** | **28,834** | **37%** | 11 MCP tools + generated docs + rules |
+| rails-ai-context CLAUDE.md only | 33,106 | 27% | Generated docs + rules, no MCP tools |
+| Normal Claude `/init` | 40,700 | 11% | Generic CLAUDE.md only |
+| No rails-ai-context at all | 45,477 | baseline | Nothing — discovers everything from scratch |
 
-All 4 produced the same working feature. The only difference was how many tokens were burned getting there.
+```
+No rails-ai-context          45,477 tk  █████████████████████████████████████████████
+Normal Claude /init           40,700 tk  █████████████████████████████████████████     -11%
+rails-ai-context CLAUDE.md    33,106 tk  █████████████████████████████████             -27%
+rails-ai-context (full)       28,834 tk  █████████████████████████████                 -37%
+```
+
+https://github.com/user-attachments/assets/14476243-1210-4e62-9dc5-9d4aa9caef7e
+
+
+**What each layer gives you:**
+
+| | Normal `/init` | rails-ai-context CLAUDE.md | rails-ai-context full |
+|---|---|---|---|
+| Knows it's Rails + Tailwind | Yes | Yes | Yes |
+| Knows model names, columns, associations | No | Yes | Yes |
+| Knows controller actions, filters | No | Yes | Yes |
+| Discovery overhead | ~8 calls | 0 calls | 0 calls |
+| Structured MCP queries | No | No | Yes — 5 MCP calls replace file reads |
+
+**~16,600 fewer tokens per task** vs no gem at all.
+
+> **This was a simple task on a small 5-model app.** Real-world tasks are 3-10x more complex.
+> A feature touching auth + payments + mailers + tests on a 50-model app? Without the gem, Claude reads `db/schema.rb` (2,000+ lines), every model file, every controller, every view — easily 200K+ tokens per session. With rails-ai-context, MCP tools return only what's needed: `rails_get_schema(table:"users")` returns 25 lines instead of 2,000. **The bigger your app and the harder the task, the more you save.**
 
-https://github.com/user-attachments/assets/171f52ae-bd30-43f6-a44f-bcfdda7fc139
+| App size | Without gem | With rails-ai-context | Savings |
+|----------|-------------|----------------------|---------|
+| Small (5 models) | 45K tokens | 29K tokens | 37% |
+| Medium (30 models) | ~150K tokens | ~60K tokens | ~60% |
+| Large (100+ models) | ~500K+ tokens | ~100K tokens | ~80% |
 
-MCP tools give the AI structured, filtered access to your codebase instead of reading entire files. On this small 5-model app, that saved 35%. **On larger projects, the savings compound significantly.**
+*Medium/large estimates based on schema.rb scaling (40 lines/table), model file scaling, and MCP summary-first workflow eliminating full-file reads.*
 
 ---
 
@@ -60,10 +87,10 @@ The install generator creates `.mcp.json` for auto-discovery — Claude Code and
 
 ![Token Comparison](https://raw.githubusercontent.com/crisnahine/rails-ai-context/main/docs/token-comparison.jpeg)
 
-- Compact context files load ≤150 lines instead of thousands
-- MCP tools return `detail:"summary"` first (~55 tokens for schema overview), then drill into specifics
-- Specific lookups (`table:`, `model:`, `controller:`) return only what's needed
-- Pagination prevents dumping hundreds of tables/routes at once
+- `/init` saves 11% — knows the framework but wastes tokens discovering models and tables
+- **CLAUDE.md saves 27%** — complete Rails-specific map, zero discovery overhead
+- **Full MCP saves 37%** — structured queries replace expensive full-file reads
+- MCP tools return `detail:"summary"` first (~55 tokens), then drill into specifics
 - Split rule files only activate in relevant directories
 
 ---

data/demo_script.sh

@@ -8,7 +8,7 @@ echo 'Fetching gem metadata from https://rubygems.org...'
 sleep 0.3
 echo 'Resolving dependencies...'
 sleep 0.3
-echo 'Installing rails-ai-context 0.10.1'
+echo 'Installing rails-ai-context 0.10.2'
 echo ''
 sleep 1
 

data/lib/rails_ai_context/introspectors/model_introspector.rb

@@ -169,8 +169,8 @@ module RailsAiContext
         macros = {}
 
         macros[:has_secure_password] = true if source.match?(/\bhas_secure_password\b/)
-        macros[:encrypts] = source.scan(/\bencrypts\s+(.+)/).flat_map { |m| m[0].scan(/:(\w+)/).flatten } if source.match?(/\bencrypts\s+:/)
-        macros[:normalizes] = source.scan(/\bnormalizes\s+(.+)/).flat_map { |m| m[0].scan(/:(\w+)/).flatten } if source.match?(/\bnormalizes\s+:/)
+        macros[:encrypts] = source.scan(/\bencrypts\s+(.+?)$/).flat_map { |m| m[0].scan(/:(\w+)/).flatten } if source.match?(/\bencrypts\s+:/)
+        macros[:normalizes] = source.scan(/\bnormalizes\s+(.+?)$/).flat_map { |m| m[0].scan(/:(\w+)/).flatten } if source.match?(/\bnormalizes\s+:/)
         macros[:has_one_attached] = source.scan(/\bhas_one_attached\s+:(\w+)/).flatten if source.match?(/\bhas_one_attached\s+:/)
         macros[:has_many_attached] = source.scan(/\bhas_many_attached\s+:(\w+)/).flatten if source.match?(/\bhas_many_attached\s+:/)
         macros[:has_rich_text] = source.scan(/\bhas_rich_text\s+:(\w+)/).flatten if source.match?(/\bhas_rich_text\s+:/)
@@ -180,7 +180,7 @@ module RailsAiContext
         macros[:store] = source.scan(/\bstore(?:_accessor)?\s+:(\w+)/).flatten if source.match?(/\bstore(?:_accessor)?\s+:/)
 
         # Delegations
-        delegations = source.scan(/\bdelegate\s+(.+?),\s*to:\s*:(\w+)/).map do |methods_str, target|
+        delegations = source.scan(/\bdelegate\s+(.+?),\s*to:\s*:(\w+)/m).map do |methods_str, target|
           { methods: methods_str.scan(/:(\w+)/).flatten, to: target }
         end
         macros[:delegations] = delegations if delegations.any?

data/lib/rails_ai_context/tools/get_view.rb

@@ -92,6 +92,8 @@ module RailsAiContext
         end
       end
 
+      MAX_FILE_SIZE = 2_000_000 # 2MB safety limit
+
       private_class_method def self.read_view_file(path)
         views_dir = Rails.root.join("app", "views")
         full_path = views_dir.join(path)
@@ -103,6 +105,9 @@ module RailsAiContext
         unless File.exist?(full_path)
           return text_response("View not found: #{path}")
         end
+        if File.size(full_path) > MAX_FILE_SIZE
+          return text_response("File too large: #{path} (#{File.size(full_path)} bytes)")
+        end
 
         content = File.read(full_path)
         text_response("# #{path}\n\n```erb\n#{content}\n```")

data/lib/rails_ai_context/tools/search_code.rb

@@ -106,7 +106,11 @@ module RailsAiContext
 
       private_class_method def self.search_with_ruby(pattern, search_path, file_type, max_results, root)
         results = []
-        regex = Regexp.new(pattern, Regexp::IGNORECASE)
+        begin
+          regex = Regexp.new(pattern, Regexp::IGNORECASE, timeout: 2)
+        rescue RegexpError => e
+          return [ { file: "error", line_number: 0, content: "Invalid regex: #{e.message}" } ]
+        end
         glob = file_type ? "**/*.#{file_type}" : "**/*.{rb,js,erb,yml,yaml,json}"
         excluded = RailsAiContext.configuration.excluded_paths
 

data/lib/rails_ai_context/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module RailsAiContext
-  VERSION = "0.10.1"
+  VERSION = "0.10.2"
 end

data/server.json

@@ -7,11 +7,11 @@
     "url": "https://github.com/crisnahine/rails-ai-context",
     "source": "github"
   },
-  "version": "0.10.1",
+  "version": "0.10.2",
   "packages": [
     {
       "registryType": "mcpb",
-      "identifier": "https://github.com/crisnahine/rails-ai-context/releases/download/v0.10.1/rails-ai-context-mcp.mcpb",
+      "identifier": "https://github.com/crisnahine/rails-ai-context/releases/download/v0.10.2/rails-ai-context-mcp.mcpb",
       "fileSha256": "dd711a0ad6c4de943ae4da94eaf59a6dc9494b9d57f726e24649ed4e2f156990",
       "transport": {
         "type": "stdio"

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: rails-ai-context
 version: !ruby/object:Gem::Version
-  version: 0.10.1
+  version: 0.10.2
 platform: ruby
 authors:
 - crisnahine