second-opinion: use Codex CLI's built-in MCP server

.claude-plugin/marketplace.json

@@ -245,8 +245,8 @@
     },
     {
       "name": "second-opinion",
-      "version": "1.5.1",
-      "description": "Runs code reviews using external LLM CLIs (OpenAI Codex, Google Gemini) on uncommitted changes, branch diffs, or specific commits. Bundles codex-mcp-server for direct MCP tool access to Codex.",
+      "version": "1.6.0",
+      "description": "Runs code reviews using external LLM CLIs (OpenAI Codex, Google Gemini) on uncommitted changes, branch diffs, or specific commits. Bundles Codex's built-in MCP server for direct tool access.",
       "author": {
         "name": "Dan Guido"
       },

.gitignore

@@ -13,6 +13,7 @@ __pycache__/
 venv/
 ENV/
 .uv/
+uv.lock
 
 # IDE
 .idea/
@@ -31,5 +32,5 @@ build/
 
 # MCP server config (contains API keys)
 .mcp.json
-# Allow plugin-bundled MCP configs (no secrets, just uvx commands)
+# Allow plugin-bundled MCP configs (no secrets, just CLI commands)
 !plugins/**/.mcp.json

README.md

@@ -91,7 +91,7 @@ cd /path/to/parent  # e.g., if repo is at ~/projects/skills, be in ~/projects
 | [let-fate-decide](plugins/let-fate-decide/) | Draw Tarot cards using cryptographic randomness to add entropy to vague planning |
 | [modern-python](plugins/modern-python/) | Modern Python tooling and best practices with uv, ruff, and pytest |
 | [seatbelt-sandboxer](plugins/seatbelt-sandboxer/) | Generate minimal macOS Seatbelt sandbox configurations |
-| [second-opinion](plugins/second-opinion/) | Run code reviews using external LLM CLIs (OpenAI Codex, Google Gemini) on changes, diffs, or commits. Bundles codex-mcp-server. |
+| [second-opinion](plugins/second-opinion/) | Run code reviews using external LLM CLIs (OpenAI Codex, Google Gemini) on changes, diffs, or commits. Bundles Codex's built-in MCP server. |
 | [skill-improver](plugins/skill-improver/) | Iterative skill refinement loop using automated fix-review cycles |
 | [workflow-skill-design](plugins/workflow-skill-design/) | Design patterns for workflow-based Claude Code skills with review agent |
 

plugins/second-opinion/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "second-opinion",
-  "version": "1.5.1",
-  "description": "Runs code reviews using external LLM CLIs (OpenAI Codex, Google Gemini) on uncommitted changes, branch diffs, or specific commits. Bundles codex-mcp-server for direct MCP tool access to Codex.",
+  "version": "1.6.0",
+  "description": "Runs code reviews using external LLM CLIs (OpenAI Codex, Google Gemini) on uncommitted changes, branch diffs, or specific commits. Bundles Codex's built-in MCP server for direct tool access.",
   "author": {
     "name": "Dan Guido"
   }

plugins/second-opinion/.mcp.json

@@ -1,8 +1,8 @@
 {
   "mcpServers": {
-    "codex-mcp-server": {
-      "command": "uvx",
-      "args": ["codex-mcp-server"]
+    "codex": {
+      "command": "codex",
+      "args": ["mcp-server"]
     }
   }
 }

plugins/second-opinion/README.md

@@ -50,10 +50,9 @@ Shells out to `codex review` and/or `gemini` CLI with high-capability model conf
 
 ## Codex MCP Tools
 
-This plugin bundles a [codex-mcp-server](https://github.com/trailofbits/codex-mcp) that auto-starts when the plugin is installed, providing three MCP tools:
+This plugin bundles Codex CLI's built-in MCP server (`codex mcp-server`), which auto-starts when the plugin is installed and provides two MCP tools:
 
-- **codex_ask** — ask Codex a question inline, swapping in an OpenAI model mid-session
-- **codex_exec** — send a raw prompt to Codex with full control over the input
-- **codex_review** — review a git diff for bugs, security issues, and correctness
+- **codex** — start a new Codex session with a prompt, model, sandbox, and approval policy settings
+- **codex-reply** — continue an existing session by thread ID for multi-turn conversations
 
 These tools work independently of the `/second-opinion` slash command. Use them when you want direct, programmatic access to Codex without the interactive prompt workflow.