@enfyra/mcp-server 0.0.81 → 0.0.83

package/README.md

@@ -12,7 +12,7 @@ From your project root:
 npx @enfyra/mcp-server config
 ```
 
-The config command can write project config for Codex, Claude Code, Cursor, or all supported clients. It preserves other MCP servers and replaces only the `enfyra` entry.
+The config command can write user-level Codex config plus project config for Claude Code and Cursor. It preserves other MCP servers and replaces only the `enfyra` entry.
 
 Interactive setup asks for `ENFYRA_APP_URL` first, then asks for `ENFYRA_API_TOKEN`. The MCP API base is inferred as `<app>/api`, and the token page is inferred as `<app>/me`. For example, `https://demo.enfyra.io` becomes API base `https://demo.enfyra.io/api` and token page `https://demo.enfyra.io/me`.
 
@@ -37,7 +37,7 @@ yarn mcp:config
 
 | Client | Command | Project config |
 |--------|---------|----------------|
-| Codex | `npx @enfyra/mcp-server config --codex` | `.codex/config.toml` |
+| Codex | `npx @enfyra/mcp-server config --codex` | `~/.codex/config.toml` |
 | Claude Code | `npx @enfyra/mcp-server config --claude-code` | `.mcp.json` |
 | Cursor | `npx @enfyra/mcp-server config --cursor` | `.cursor/mcp.json` |
 | MCP Inspector / other hosts | Paste the shared STDIO config below | Host-specific `mcpServers` config |
@@ -49,7 +49,7 @@ yarn mcp:config
 npx @enfyra/mcp-server config --codex
 ```
 
-Generated project config:
+Generated user config:
 
 ```toml
 [mcp_servers.enfyra]
@@ -61,9 +61,9 @@ ENFYRA_API_URL = "http://localhost:3000/api"
 ENFYRA_API_TOKEN = "efy_pat_your-token"
 ```
 
-The writer replaces only `[mcp_servers.enfyra]` and `[mcp_servers.enfyra.env]`. Other Codex config and other MCP servers are preserved.
+The writer replaces only `[mcp_servers.enfyra]` and `[mcp_servers.enfyra.env]` in the user-level Codex config. Other Codex config and other MCP servers are preserved.
 
-After editing config, open the folder in a new Codex session so project config is loaded.
+Restart Codex or start a new Codex session so user MCP config is loaded. The setup command does not create project `.codex/config.toml`, because project-scoped executable MCP config can trigger Codex workspace safety checks.
 
 Official reference: [Codex config](https://developers.openai.com/codex/config-reference).
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@enfyra/mcp-server",
-  "version": "0.0.81",
+  "version": "0.0.83",
   "description": "MCP server for Enfyra - manage Enfyra instances from MCP-compatible coding tools",
   "type": "module",
   "license": "MIT",
@@ -10,8 +10,7 @@
     "enfyra-mcp-server": "src/index.mjs"
   },
   "files": [
-    "src",
-    ".codex/skills"
+    "src"
   ],
   "scripts": {
     "start": "node src/index.mjs",

package/src/lib/config-local.mjs

@@ -23,7 +23,7 @@ const style = {
 const clients = {
   codex: {
     label: 'Codex',
-    path: './.codex/config.toml',
+    path: '~/.codex/config.toml',
     color: style.green,
   },
   claude: {
@@ -61,7 +61,7 @@ ${style.bold('Usage')}
   npx @enfyra/mcp-server config [options]
 
 ${style.bold('Supported clients')}
-  Codex        ./.codex/config.toml
+  Codex        ~/.codex/config.toml
   Claude Code  ./.mcp.json
   Cursor       ./.cursor/mcp.json
   Other MCP hosts can use the shared stdio JSON from the README.
@@ -69,7 +69,7 @@ ${style.bold('Supported clients')}
 ${style.bold('Options')}
   --app-url <url>          Enfyra app/admin URL, for example https://demo.enfyra.io
   --api-token, -t <secret>  ENFYRA_API_TOKEN
-  --global                Write global/user config for selected hosts instead of project config
+  --global                Write Claude Code/Cursor global config instead of project config; Codex is always user-level
   --reconfig              Always choose target again in interactive mode and replace the old enfyra config for that target
   --yes                   Non-interactive: no prompts (CI / scripts); use CLI, env, existing file, then defaults
 
@@ -78,7 +78,7 @@ ${style.bold('Client selection')}
 
   --claude-code, --claude, --claude-only   Only ./.mcp.json (Claude Code project scope)
   --cursor, --cursor-only                  Only ./.cursor/mcp.json (Cursor)
-  --codex, --codex-only                    Only ./.codex/config.toml (Codex project scope)
+  --codex, --codex-only                    Only ~/.codex/config.toml (Codex user scope)
   Passing multiple target flags writes each selected target.
 
   -h, --help              Show this help
@@ -287,7 +287,7 @@ async function readCodexEnfyraEnv(absPath) {
 }
 
 function getCodexConfigPath(root, globalScope) {
-  return globalScope ? join(homedir(), '.codex', 'config.toml') : join(root, '.codex', 'config.toml');
+  return join(homedir(), '.codex', 'config.toml');
 }
 
 function getClaudeConfigPath(root, globalScope) {
@@ -361,7 +361,7 @@ async function promptTargetChoice() {
   const rl = createInterface({ input, output });
   const line = (await rl.question(
     'Where should Enfyra MCP config be written?\n'
-      + '  [1] Codex        ./.codex/config.toml\n'
+      + '  [1] Codex        ~/.codex/config.toml\n'
       + '  [2] Claude Code  ./.mcp.json\n'
       + '  [3] Cursor       ./.cursor/mcp.json\n'
       + '  [4] All [default]\n'
@@ -595,7 +595,14 @@ export async function runLocalConfig(argv) {
     written.push({ client: 'cursor', path: p });
   }
 
-  const scopeLabel = opts.global ? 'global/user' : 'project';
+  const hasProjectConfig = writeClaude || writeCursor;
+  const scopeLabel = opts.global
+    ? 'global/user'
+    : writeCodex && hasProjectConfig
+      ? 'project + Codex user'
+      : writeCodex
+        ? 'Codex user'
+        : 'project';
   console.log(`${statusIcon('success')} ${style.bold(style.green('Enfyra MCP config updated'))} ${style.dim(`(${scopeLabel})`)}\n`);
   for (const entry of written) {
     const meta = clients[entry.client];
@@ -606,7 +613,7 @@ export async function runLocalConfig(argv) {
   const selectedClients = new Set(written.map(entry => entry.client));
   console.log(`\n${style.bold(style.blue('Next steps'))}`);
   if (selectedClients.has('codex')) {
-    console.log('  - Codex: open this folder in a new Codex session so config is loaded.');
+    console.log('  - Codex: restart Codex or start a new Codex session so user MCP config is loaded.');
   }
   if (selectedClients.has('claude')) {
     console.log('  - Claude Code: open this folder; approve project MCP if prompted.');

package/.codex/skills/enfyra-mcp-performance/SKILL.md

@@ -1,47 +0,0 @@
-# Enfyra MCP Performance And Debug Mode
-
-Use this skill when designing, reviewing, or debugging Enfyra apps through MCP where performance, read/write shape, indexes, websocket latency, RLS filters, or query correctness matter.
-
-## Rules
-- Verify capability with live metadata before claiming performance behavior. Use `inspect_table`, `inspect_route`, `discover_query_capabilities`, and `discover_runtime_context` first.
-- Prefer indexed relation filters over scalar mirror columns. Relation property names can be used in `indexes` and `uniques`; Enfyra resolves them to physical FK columns.
-- For hot read paths, design indexes with the most selective/user-scoped field first. Examples: `["member","is_read","conversation"]` for unread lookup and `["conversation","member","is_read"]` for mark-read updates.
-- Use existence checks for UI dots and badges unless the user explicitly needs exact counts. Avoid count queries on every conversation row.
-- Use `meta=filterCount` or MCP `count_records` only when count is the product requirement.
-- For RLS hooks, mutate `@QUERY.filter` and preserve existing user filters with `_and`. `@QUERY.filter` is already `{}` when omitted.
-- For dynamic scripts, keep runtime values in their original type. Do not wrap ids or payload values in `String(...)`, `Number(...)`, or `Boolean(...)`.
-- For websocket apps, connect browsers through the Enfyra app/Nuxt bridge, not the hidden backend. Event handlers should be script-owned and use `@SOCKET` explicitly.
-- Authenticated websocket gateways load `user_definition` once and expose it as `@USER`; do not ask clients to send their own `senderId`.
-- Enfyra automatically joins authenticated sockets to `user_<userId>` after the connection script succeeds. App scripts should use this for `emitToUser` and should not re-join that room manually.
-
-## Debug Workflow
-1. Inspect the table schema, relations, indexes, and route permissions before changing code.
-2. Reproduce with the smallest real request or Socket.IO event. Prefer `test_rest_endpoint` or `run_admin_test` when available.
-3. If the problem is performance, state the exact query shape and expected index. Add or update `indexes` on `table_definition` through `create_table` or `update_table`.
-4. Reload metadata/cache after schema or script changes when the tool does not do it automatically.
-5. Retest the exact route/event and compare behavior before/after. Do not invent benchmark numbers.
-
-## Chat App Review Checklist
-- Confirm REST and Socket.IO both go through the app origin. REST uses the app proxy prefix; Socket.IO uses `/ws/<namespace>` and `path: "/ws/socket.io"` when the app bridge exposes that path.
-- Confirm browser code never stores or forwards custom JWT cookies when the Enfyra app/proxy already manages cookies.
-- Confirm `chat:join` queries `chat_conversation` visible to `@USER`, then joins `conversation:<id>`. Do not join rooms from raw membership member ids.
-- Confirm `chat:message` uses `@SOCKET.broadcastToRoom("conversation:" + conversationId, ...)` and persists with `@REPOS` inside the event script. Do not add a flow just to save chat messages.
-- Confirm new DM UX creates no empty conversation. A draft opens first; the first message calls a creation event that creates the conversation and message together.
-- Confirm route-level RLS is server-side: pre-hooks merge membership filters into `@QUERY.filter` with `_and`; the client must not fetch all conversations then filter locally.
-- Confirm conversation titles are computed from visible memberships from the current user's perspective. DM title should be the other person, not the current user.
-- Confirm message history uses cursor pagination, newest messages first from the API then reversed for display; loading older messages preserves scroll.
-- Confirm disconnect state disables chat input and shows a visible retry banner immediately.
-- Confirm typing state is room-scoped, user-aware, and remains active while the input has text, even if the user pauses typing.
-
-## Chat Read/Unread Pattern
-- Use a join table, e.g. `chat_message_read`, with:
-  - `message` relation to `chat_message`
-  - `conversation` relation to `chat_conversation`
-  - `member` relation to `user_definition`
-  - `is_read` boolean
-  - `read_at` date nullable
-- Add unique `["message","member"]`.
-- Add indexes `["member","is_read","conversation"]` and `["conversation","member","is_read"]`.
-- On message create, create read rows for conversation members. Sender rows start as read; other member rows start unread.
-- On conversation open/read, update unread rows for `@USER` in that conversation to read.
-- UI unread dots should check whether an unread row exists; count only when requested.