rails-ai-context 0.7.1 → 0.8.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7aa2c43d22465356700ded7ac615f4d38ba24e1bee512658dc1ec43196935e9c
-  data.tar.gz: f718d8e2c3b790294e212c557216cd2babd99459fea3a5f1245f9ad09329e52d
+  metadata.gz: 04212baad7453e8371662a226fa50dfa4b65e4b611e94af22c50ab09e314d888
+  data.tar.gz: 43140d9b9f6151a4f5e72427efd8fcba0ed4a8256ed3f3c56f6a1c1ff53fd436
 SHA512:
-  metadata.gz: c4c9d908941193ffd7cda89bc8a5da67a9531ffd09e723382eebc39c7ac77a06faa3958bf4048f550d0d3a5883edf9f8a67a2b62c65130847589db53d954fe5d
-  data.tar.gz: 38a1f37c770eff7680ed3b0acb43627a5ba36b8168eafbf087907d8e5445103ec4cc06b4cb08103e48baf772c710d6333278759789f3bcc44e7b745787cc9b11
+  metadata.gz: 373de7c5e83e959cc64c96ab3937e9160da315c9dde34e12f2fe506f5ee6b815cbe71a49af3e6015fe0b75b89316810ff067deb0fa839a5dd4cff6836cb33f07
+  data.tar.gz: 8c9b16584f464c0603cb04f92ab8d7daf0f8dfe04f15a78e21a6089c344bcd0aedf9a6a59e748e9a90d24a356ad27f0b69cde8512aa3fd7db1995a84641e7a22

data/CHANGELOG.md

@@ -5,11 +5,30 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.8.1] - 2026-03-19
+
+### Added
+
+- **OpenCode support** — generates `AGENTS.md` (native OpenCode context file) plus per-directory `app/models/AGENTS.md` and `app/controllers/AGENTS.md` that OpenCode auto-loads when reading files in those directories. Falls back to `CLAUDE.md` when no `AGENTS.md` exists. New command: `rails ai:context:opencode`.
+
+### Fixed
+
+- **Live reload LoadError in HTTP mode** — when `live_reload = true` and the `listen` gem was missing, the `start_http` method's rescue block (for rackup fallback) swallowed the live reload error, producing a confusing rack error instead of the correct "listen gem required" message. The rescue is now scoped to the rackup require only.
+- **Dangling @live_reload reference** — `@live_reload` was assigned before `start` was called. If `start` raised LoadError, the instance variable pointed to a non-functional object. Now only assigned after successful start.
+
+## [0.8.0] - 2026-03-19
+
+### Added
+
+- **MCP Live Reload** — when running `rails ai:serve`, file changes automatically invalidate tool caches and send MCP notifications (`notifications/resources/list_changed`) to connected AI clients. The AI's context stays fresh without manual re-querying. Requires the `listen` gem (enabled by default when available). Configurable via `config.live_reload` (`:auto`, `true`, `false`) and `config.live_reload_debounce` (default: 1.5s).
+- **Live reload doctor check** — `rails ai:doctor` now warns when the `listen` gem is not installed.
+
 ## [0.7.1] - 2026-03-19
 
 ### Added
 
 - **Full MCP tool reference in all context files** — every generated file (CLAUDE.md, .cursorrules, .windsurfrules, copilot-instructions.md) now includes complete tool documentation with parameters, detail levels, pagination examples, and usage workflow. Dedicated `rails-mcp-tools` split rule files added for Claude, Cursor, Windsurf, and Copilot.
+- **MCP Registry listing** — published to the [official MCP Registry](https://registry.modelcontextprotocol.io) as `io.github.crisnahine/rails-ai-context` via mcpb package type.
 
 ### Fixed
 

data/CLAUDE.md

@@ -10,12 +10,13 @@ structure to AI assistants via the Model Context Protocol (MCP).
 - `lib/rails_ai_context/introspector.rb` — Orchestrates sub-introspectors
 - `lib/rails_ai_context/introspectors/` — 27 introspectors (schema, models, routes, jobs, gems, conventions, stimulus, database_stats, controllers, views, turbo, i18n, config, active_storage, action_text, auth, api, tests, rake_tasks, assets, devops, action_mailbox, migrations, seeds, middleware, engines, multi_database)
 - `lib/rails_ai_context/tools/` — 9 MCP tools using the official mcp SDK
-- `lib/rails_ai_context/serializers/` — Output formatters (claude, claude_rules, cursor_rules, windsurf, windsurf_rules, copilot, copilot_instructions, rules, markdown, JSON)
+- `lib/rails_ai_context/serializers/` — Output formatters (claude, claude_rules, opencode, cursor_rules, windsurf, windsurf_rules, copilot, copilot_instructions, rules, markdown, JSON)
 - `lib/rails_ai_context/resources.rb` — MCP resources (static data AI clients read directly)
 - `lib/rails_ai_context/server.rb` — MCP server configuration (stdio + HTTP transports)
 - `lib/rails_ai_context/middleware.rb` — Rack middleware for auto-mounting MCP HTTP endpoint
 - `lib/rails_ai_context/fingerprinter.rb` — SHA256 file fingerprinting for cache invalidation
 - `lib/rails_ai_context/doctor.rb` — Diagnostic checks and AI readiness scoring
+- `lib/rails_ai_context/live_reload.rb` — MCP live reload: watches files, invalidates caches, notifies AI clients
 - `lib/rails_ai_context/watcher.rb` — File watcher for auto-regenerating context files
 - `lib/rails_ai_context/engine.rb` — Rails Engine for auto-integration
 - `lib/generators/rails_ai_context/install/` — Install generator (creates .mcp.json, initializer, context files)
@@ -38,7 +39,7 @@ structure to AI assistants via the Model Context Protocol (MCP).
 ## Testing
 
 ```bash
-bundle exec rspec           # Run specs (348 examples)
+bundle exec rspec           # Run specs (373 examples)
 bundle exec rubocop         # Lint
 ```
 

data/CONTRIBUTING.md

@@ -20,8 +20,9 @@ The test suite uses [Combustion](https://github.com/pat/combustion) to boot a mi
 lib/rails_ai_context/
 ├── introspectors/     # 27 introspectors (schema, models, routes, etc.)
 ├── tools/             # 9 MCP tools with detail levels and pagination
-├── serializers/       # Per-assistant formatters (claude, cursor, windsurf, copilot, JSON)
+├── serializers/       # Per-assistant formatters (claude, opencode, cursor, windsurf, copilot, JSON)
 ├── server.rb          # MCP server setup (stdio + HTTP)
+├── live_reload.rb     # MCP live reload (file watcher + cache invalidation)
 ├── engine.rb          # Rails Engine for auto-integration
 └── configuration.rb   # User-facing config (presets, context_mode, limits)
 ```