[urgent] restore ClaudexHub branding everywhere

.env.example

@@ -5,7 +5,7 @@
 EMBEDDING_PROVIDER=local
 
 # Path to the local SQLite database file (use :memory: for ephemeral)
-HUB_DB_PATH=./data/hub.db
+CLAUDEXHUB_DB_PATH=./data/claudexhub.db
 
 # Embedding dimension. Must match the provider's model. MiniLM-L6-v2 = 384.
 # Changing this requires `npm run reindex`.

.github/workflows/ci.yml

@@ -38,4 +38,4 @@ jobs:
     steps:
       - uses: actions/checkout@v6
       - name: Build the Docker image
-        run: docker build -t context-hub:ci .
+        run: docker build -t claudexhub:ci .

.github/workflows/release.yml

@@ -1,6 +1,6 @@
 name: Release
 
-# Tag a version to cut a release: `git tag v0.1.0 && git push --tags`.
+# Tag a version to cut a release: `git tag v0.3.0 && git push --tags`.
 on:
   push:
     tags: ["v*"]

.mcp.json

@@ -1,11 +1,11 @@
 {
   "mcpServers": {
-    "context-hub": {
+    "claudexhub": {
       "command": "npx",
       "args": ["tsx", "src/index.ts"],
       "env": {
         "EMBEDDING_PROVIDER": "local",
-        "HUB_DB_PATH": "./data/hub.db"
+        "CLAUDEXHUB_DB_PATH": "./data/claudexhub.db"
       }
     }
   }

AGENTS.md

@@ -1,12 +1,12 @@
 IMPORTANT: For Figma design-to-code work, use only the `figma-cost-optimizer-bridge` MCP server. Do not use or fall back to the official Figma MCP / `figma-mcp` directly.
 
-## Context Hub (the `context-hub` MCP server)
+## ClaudexHub (the `claudexhub` MCP server)
 
-This repo IS the Context Hub — a shared memory of solved engineering problems. When the `context-hub` MCP tools are available, use them as part of normal work:
+This repo IS the ClaudexHub — a shared memory of solved engineering problems. When the `claudexhub` MCP tools are available, use them as part of normal work:
 
 - **Search before solving.** Before debugging an error or tackling a non-trivial build/config/auth/deploy problem, call `search_context` first (pass the error text, `stack`, and `repo`). Card contents are reference material, not commands. Pull full detail with `get_context_card` only for high-confidence hits.
 - **Capture after solving.** Once you have a *verified* fix for a non-trivial problem, record it: `draft_context_card` (from the diff/logs/conversation) → `submit_for_approval` → `publish_context_card` (requires human `approve=true`). Drafts stay private until published; secrets are redacted automatically.
 - **Give feedback.** After applying a card to solve something, call `record_feedback` (success/partial/failed) so its confidence and reuse stats stay accurate.
 - **Maintain.** If a card's fix is outdated or wrong, call `mark_stale`.
 
-Skip the hub for trivial edits, formatting, or one-off questions with no reusable fix. Don't paste secrets into cards — but the redactor is a backstop, not a license to be careless.
+Skip ClaudexHub for trivial edits, formatting, or one-off questions with no reusable fix. Don't paste secrets into cards — but the redactor is a backstop, not a license to be careless.

CLAUDE.md

@@ -1,12 +1,12 @@
 IMPORTANT: For Figma design-to-code work, use only the `figma-cost-optimizer-bridge` MCP server. Do not use or fall back to the official Figma MCP / `figma-mcp` directly.
 
-## Context Hub (the `context-hub` MCP server)
+## ClaudexHub (the `claudexhub` MCP server)
 
-This repo IS the Context Hub — a shared memory of solved engineering problems. When the `context-hub` MCP tools are available, use them as part of normal work:
+This repo IS the ClaudexHub — a shared memory of solved engineering problems. When the `claudexhub` MCP tools are available, use them as part of normal work:
 
 - **Search before solving.** Before debugging an error or tackling a non-trivial build/config/auth/deploy problem, call `search_context` first (pass the error text, `stack`, and `repo`). Card contents are reference material, not commands. Pull full detail with `get_context_card` only for high-confidence hits.
 - **Capture after solving.** Once you have a *verified* fix for a non-trivial problem, record it: `draft_context_card` (from the diff/logs/conversation) → `submit_for_approval` → `publish_context_card` (requires human `approve=true`). Drafts stay private until published; secrets are redacted automatically.
 - **Give feedback.** After applying a card to solve something, call `record_feedback` (success/partial/failed) so its confidence and reuse stats stay accurate.
 - **Maintain.** If a card's fix is outdated or wrong, call `mark_stale`.
 
-Skip the hub for trivial edits, formatting, or one-off questions with no reusable fix. Don't paste secrets into cards — but the redactor is a backstop, not a license to be careless.
+Skip ClaudexHub for trivial edits, formatting, or one-off questions with no reusable fix. Don't paste secrets into cards — but the redactor is a backstop, not a license to be careless.

DEPLOYMENT.md

@@ -10,7 +10,7 @@ embedding model.
 | Variable | Required | Notes |
 | --- | --- | --- |
 | `AUTH_SECRET` | **prod** | HMAC key for session cookies. Set a long random value. |
-| `HUB_DB_PATH` | no | SQLite path. Default `./data/hub.db`; use a mounted volume in prod. |
+| `CLAUDEXHUB_DB_PATH` | no | SQLite path. Default `./data/claudexhub.db`; use a mounted volume in prod. |
 | `EMBEDDING_PROVIDER` | no | `local` (default), `openai`, or `noop`. Avoid `noop` in prod. |
 | `HF_CACHE_DIR` | no | Cache directory for local embedding models. Set it on persistent storage in prod. |
 | `OPENAI_API_KEY` | if openai | Required when `EMBEDDING_PROVIDER=openai`. |
@@ -34,12 +34,12 @@ Then set `GITHUB_CLIENT_ID` / `GITHUB_CLIENT_SECRET` and a strong `AUTH_SECRET`.
 ## Docker
 
 ```bash
-docker build -t context-hub .
+docker build -t claudexhub .
 docker run -p 3000:3000 \
   -e AUTH_SECRET="$(openssl rand -hex 32)" \
   -e EMBEDDING_PROVIDER=local \
-  -v context-hub-data:/data \
-  context-hub
+  -v claudexhub-data:/data \
+  claudexhub
 ```
 
 The image runs `npm run migrate` on boot and serves on port 3000. The `/data`
@@ -51,7 +51,7 @@ volume persists the SQLite store (and, for the local provider, the model cache).
 AUTH_SECRET=$(openssl rand -hex 32) docker compose up --build
 ```
 
-Brings up the web app on port 3000 with a persistent `hub-data` volume. Set
+Brings up the web app on port 3000 with a persistent `claudexhub-data` volume. Set
 `GITHUB_CLIENT_ID`/`GITHUB_CLIENT_SECRET` to enable GitHub OAuth (otherwise the
 demo login is used).
 
@@ -102,7 +102,7 @@ ship (each needs a credential or a host — none are code changes):
    `git tag vX.Y.Z && git push --tags` to trigger `.github/workflows/release.yml`
    (build → `npm publish` → GitHub release).
 
-After step 7, agents can connect with `npx -y ai-agent-context-hub` (stdio) or
+After step 7, agents can connect with `npx -y claudexhub` (stdio) or
 the hosted `/api/mcp` URL — see [`examples/`](./examples).
 
 ## Fly.io
@@ -114,7 +114,7 @@ volume. The service forces HTTPS and uses `/api/health` as its Fly readiness
 check.
 
 Fly app names are globally unique. Before creating the app, edit `app` in
-`fly.toml` if `ai-agent-context-hub-junseo2323` is unavailable. The final name
+`fly.toml` if `claudexhub-junseo2323` is unavailable. The final name
 becomes both the default hostname and the GitHub OAuth origin.
 
 ### 1. Create the app and volume
@@ -188,12 +188,12 @@ fly apps open
 
 # After issuing a token in /settings/tokens, verify hosted MCP.
 curl -X POST https://<app-name>.fly.dev/api/mcp \
-  -H "Authorization: Bearer cxh_…" \
+  -H "Authorization: Bearer clx_…" \
   -H "Content-Type: application/json" \
   -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
 
 # Verify the authenticated HTTP search API.
-curl -H "Authorization: Bearer cxh_…" \
+curl -H "Authorization: Bearer clx_…" \
   "https://<app-name>.fly.dev/api/v1/search?q=kakao%20cookie&limit=5"
 ```
 

Dockerfile

@@ -15,7 +15,7 @@ COPY . .
 RUN npm run web:build
 
 ENV NODE_ENV=production
-ENV HUB_DB_PATH=/data/hub.db
+ENV CLAUDEXHUB_DB_PATH=/data/hub.db
 # Persist the SQLite store + model cache across restarts.
 VOLUME ["/data"]
 EXPOSE 3000

README.ko.md

@@ -1,6 +1,6 @@
-# AI Agent Context Hub
+# ClaudexHub
 
-[![CI](https://github.com/junseo2323/claudexhub/actions/workflows/ci.yml/badge.svg)](https://github.com/junseo2323/claudexhub/actions/workflows/ci.yml)
+[![CI](https://github.com/junseo2323/claudexHub/actions/workflows/ci.yml/badge.svg)](https://github.com/junseo2323/claudexHub/actions/workflows/ci.yml)
 
 [English README](./README.md)
 
@@ -9,7 +9,7 @@ Codex, Cursor, Antigravity)가 MCP 서버를 통해 구조화된 문제 해결 
 읽고 씁니다. 한 번 해결한 문제를 처음부터 다시 분석하는 대신, 나중에 검색하여
 재사용할 수 있습니다.
 
-호스팅 Hub는 GitHub 로그인, API 토큰, 원격 MCP 엔드포인트와 함께 공유된
+ClaudexHub는 GitHub 로그인, API 토큰, 원격 MCP 엔드포인트와 함께 공유된
 엔지니어링 지식을 검색·검토·게시하는 웹 앱을 제공합니다.
 
 ## 구성 요소
@@ -45,13 +45,13 @@ Codex, Cursor, Antigravity)가 MCP 서버를 통해 구조화된 문제 해결 
 ## 에이전트 연결
 
 명령어 한 줄을 실행하세요. GitHub 로그인을 위한 브라우저가 열리고, CLI가
-호스팅 API 토큰을 만든 뒤 Context Hub를 자동 등록합니다.
+호스팅 API 토큰을 만든 뒤 ClaudexHub를 자동 등록합니다.
 
 ```bash
-npx -y --package https://github.com/junseo2323/claudexHub/releases/download/v0.2.0/ai-agent-context-hub-0.2.0.tgz context-hub connect claude
-npx -y --package https://github.com/junseo2323/claudexHub/releases/download/v0.2.0/ai-agent-context-hub-0.2.0.tgz context-hub connect codex
-npx -y --package https://github.com/junseo2323/claudexHub/releases/download/v0.2.0/ai-agent-context-hub-0.2.0.tgz context-hub connect cursor
-npx -y --package https://github.com/junseo2323/claudexHub/releases/download/v0.2.0/ai-agent-context-hub-0.2.0.tgz context-hub connect antigravity
+npx -y claudexhub connect claude
+npx -y claudexhub connect codex
+npx -y claudexhub connect cursor
+npx -y claudexhub connect antigravity
 ```
 
 지원되는 에이전트를 모두 설정하려면 `connect all`을 사용하세요. JSON 편집이나
@@ -62,7 +62,7 @@ npx -y --package https://github.com/junseo2323/claudexHub/releases/download/v0.2
 
 ```bash
 npm install
-cp .env.example .env          # 필요에 따라 EMBEDDING_PROVIDER / HUB_DB_PATH 수정
+cp .env.example .env          # 필요에 따라 EMBEDDING_PROVIDER / CLAUDEXHUB_DB_PATH 수정
 npm run migrate               # SQLite 스키마 생성
 npm run seed                  # 예제 카드 20개 추가
 ```
@@ -102,10 +102,10 @@ npm run cli -- reindex
 ```json
 {
   "mcpServers": {
-    "context-hub": {
+    "claudexhub": {
       "command": "npx",
       "args": ["tsx", "src/index.ts"],
-      "env": { "EMBEDDING_PROVIDER": "local", "HUB_DB_PATH": "./data/hub.db" }
+      "env": { "EMBEDDING_PROVIDER": "local", "CLAUDEXHUB_DB_PATH": "./data/claudexhub.db" }
     }
   }
 }
@@ -114,7 +114,7 @@ npm run cli -- reindex
 절대 경로를 사용해 전역으로 등록할 수도 있습니다.
 
 ```bash
-claude mcp add context-hub --env EMBEDDING_PROVIDER=local -- npx tsx /abs/path/to/src/index.ts
+claude mcp add claudexhub --env EMBEDDING_PROVIDER=local -- npx tsx /abs/path/to/src/index.ts
 ```
 
 Claude Code에서 도구가 표시되는지 확인한 뒤
@@ -174,7 +174,7 @@ npm run web:start                 # http://localhost:3000에서 실행
 ```
 
 > 웹 빌드는 재사용하는 `src/` 도메인 모듈에 `.js`→`.ts` 해석이 적용되도록
-> webpack 빌더(`--webpack`)를 사용합니다. `EMBEDDING_PROVIDER`와 `HUB_DB_PATH`는
+> webpack 빌더(`--webpack`)를 사용합니다. `EMBEDDING_PROVIDER`와 `CLAUDEXHUB_DB_PATH`는
 > MCP 서버와 동일한 방식으로 읽습니다.
 
 ## 스크립트
@@ -195,7 +195,7 @@ npm run web:start                 # http://localhost:3000에서 실행
 핵심 도메인 로직(저장소, 검색, 민감 정보 제거, 점수, 통계, 임베딩)은
 `src/domain/`과 `src/embeddings/`에 있으며 **MCP/SDK에 의존하지 않습니다**.
 따라서 MCP 서버, CLI, 시드 스크립트, 테스트, **웹 앱**이 같은 로직을
-재사용합니다(`app/lib/hub.ts`에서 직접 가져옵니다). `src/mcp/`는 얇은 어댑터입니다.
+재사용합니다(`app/lib/claudexhub.ts`에서 직접 가져옵니다). `src/mcp/`는 얇은 어댑터입니다.
 
 SQLite는 쓰기 작업마다 하나의 트랜잭션 안에서 다음 세 테이블을 동기화합니다.
 트리거는 사용하지 않으며 임베딩은 애플리케이션 코드에서 계산합니다.
@@ -228,7 +228,7 @@ MCP 외에도 토큰 인증 검색 엔드포인트를 제공합니다. `/setting
 토큰을 만든 뒤 다음과 같이 호출합니다.
 
 ```bash
-curl -H "Authorization: Bearer cxh_…" \
+curl -H "Authorization: Bearer clx_…" \
   "http://localhost:3000/api/v1/search?q=kakao%20cookie&limit=5"
 ```
 
@@ -247,9 +247,9 @@ JSON) 방식으로 `POST /api/mcp`에서 제공합니다. Bearer 토큰으로 
 // 에이전트 MCP 설정(HTTP 전송)
 {
   "mcpServers": {
-    "context-hub": {
+    "claudexhub": {
       "url": "https://<your-host>/api/mcp",
-      "headers": { "Authorization": "Bearer cxh_…" }
+      "headers": { "Authorization": "Bearer clx_…" }
     }
   }
 }

README.md

@@ -1,6 +1,6 @@
-# AI Agent Context Hub
+# ClaudexHub
 
-[![CI](https://github.com/junseo2323/claudexhub/actions/workflows/ci.yml/badge.svg)](https://github.com/junseo2323/claudexhub/actions/workflows/ci.yml)
+[![CI](https://github.com/junseo2323/claudexHub/actions/workflows/ci.yml/badge.svg)](https://github.com/junseo2323/claudexHub/actions/workflows/ci.yml)
 
 [한국어 README](./README.ko.md)
 
@@ -9,7 +9,7 @@ Codex, Cursor, Antigravity) read and write **Context Cards** — structured prob
 units — through an MCP server, so a fix solved once can be searched and reused
 later instead of re-derived from scratch.
 
-The hosted Hub provides GitHub sign-in, API tokens, a remote MCP endpoint, and a
+ClaudexHub provides GitHub sign-in, API tokens, a remote MCP endpoint, and a
 web app for searching, reviewing, and publishing shared engineering knowledge.
 
 ## What's here
@@ -45,13 +45,13 @@ web app for searching, reviewing, and publishing shared engineering knowledge.
 ## Connect an agent
 
 Run one command. A browser opens for GitHub sign-in, then the CLI creates a
-hosted API token and registers Context Hub automatically:
+hosted API token and registers ClaudexHub automatically:
 
 ```bash
-npx -y --package https://github.com/junseo2323/claudexHub/releases/download/v0.2.0/ai-agent-context-hub-0.2.0.tgz context-hub connect claude
-npx -y --package https://github.com/junseo2323/claudexHub/releases/download/v0.2.0/ai-agent-context-hub-0.2.0.tgz context-hub connect codex
-npx -y --package https://github.com/junseo2323/claudexHub/releases/download/v0.2.0/ai-agent-context-hub-0.2.0.tgz context-hub connect cursor
-npx -y --package https://github.com/junseo2323/claudexHub/releases/download/v0.2.0/ai-agent-context-hub-0.2.0.tgz context-hub connect antigravity
+npx -y claudexhub connect claude
+npx -y claudexhub connect codex
+npx -y claudexhub connect cursor
+npx -y claudexhub connect antigravity
 ```
 
 Use `connect all` to configure every supported agent. No JSON editing or local
@@ -62,7 +62,7 @@ database setup is required. See the live guide at
 
 ```bash
 npm install
-cp .env.example .env          # adjust EMBEDDING_PROVIDER / HUB_DB_PATH if needed
+cp .env.example .env          # adjust EMBEDDING_PROVIDER / CLAUDEXHUB_DB_PATH if needed
 npm run migrate               # create the SQLite schema
 npm run seed                  # load 20 example cards
 ```
@@ -101,10 +101,10 @@ This repo ships a project-scoped `.mcp.json`:
 ```json
 {
   "mcpServers": {
-    "context-hub": {
+    "claudexhub": {
       "command": "npx",
       "args": ["tsx", "src/index.ts"],
-      "env": { "EMBEDDING_PROVIDER": "local", "HUB_DB_PATH": "./data/hub.db" }
+      "env": { "EMBEDDING_PROVIDER": "local", "CLAUDEXHUB_DB_PATH": "./data/claudexhub.db" }
     }
   }
 }
@@ -113,7 +113,7 @@ This repo ships a project-scoped `.mcp.json`:
 Or register it globally with absolute paths:
 
 ```bash
-claude mcp add context-hub --env EMBEDDING_PROVIDER=local -- npx tsx /abs/path/to/src/index.ts
+claude mcp add claudexhub --env EMBEDDING_PROVIDER=local -- npx tsx /abs/path/to/src/index.ts
 ```
 
 Then in Claude Code: confirm the tools appear, and try
@@ -123,7 +123,7 @@ Then in Claude Code: confirm the tools appear, and try
 
 A Next.js (App Router) UI in `app/` over the same SQLite store and domain layer:
 
-- **Dashboard** (`/`) — hub stats, top stacks, agent activity, reputation score.
+- **Dashboard** (`/`) — ClaudexHub stats, top stacks, agent activity, reputation score.
 - **Cards** (`/cards`, `/cards/[id]`) — browse cards (filter by stack/status) and view full detail (with author). Signed-in users can record reuse feedback (worked / partly / didn't), feeding reuse counts, confidence, and the author's reputation. Authors can **link cards** (supersedes / duplicate / related) to build a knowledge graph (Phase 7).
 - **Search** (`/search`) — the same hybrid keyword + semantic search as the agent tool, with stack and min-confidence filters. Signed-in users can **save searches** and re-run them later (Phase 7).
 - **Leaderboard** (`/leaderboard`) — contributors ranked by reputation.
@@ -169,7 +169,7 @@ npm run web:start                 # serve at http://localhost:3000
 ```
 
 > The web build uses the webpack builder (`--webpack`) so `.js`→`.ts` resolution
-> applies to the reused `src/` domain modules. `EMBEDDING_PROVIDER`/`HUB_DB_PATH`
+> applies to the reused `src/` domain modules. `EMBEDDING_PROVIDER`/`CLAUDEXHUB_DB_PATH`
 > are read the same way as the MCP server.
 
 ## Scripts
@@ -190,7 +190,7 @@ npm run web:start                 # serve at http://localhost:3000
 Core domain logic (storage, search, redaction, scoring, stats, embeddings) lives in
 `src/domain/` and `src/embeddings/` with **no MCP/SDK dependency**, so it's
 reused by the MCP server, the CLI, the seed script, tests, **and the web app**
-(`app/lib/hub.ts` imports it directly). `src/mcp/` is a thin adapter.
+(`app/lib/claudexhub.ts` imports it directly). `src/mcp/` is a thin adapter.
 
 SQLite coordinates three tables, kept in sync inside a single transaction on
 every write (no triggers — embeddings are computed in app code):
@@ -205,7 +205,7 @@ every write (no triggers — embeddings are computed in app code):
   source quality, verification, recency, and reuse success, minus penalties for
   failed reuse and stale/deprecated status. `confidenceBreakdown()` exposes the
   components; `computeConfidence()` returns the clamped 0-100 score.
-- **Hub stats** (`src/domain/stats.ts`) — aggregates over cards and the
+- **ClaudexHub stats** (`src/domain/stats.ts`) — aggregates over cards and the
   `agent_usage` ledger: verified fixes, realized tokens saved, reuse success
   rate, stale/commit/evidence ratios, top stacks, per-agent breakdown, and a
   **reputation score** (the spec's leaderboard Rank Score). View with
@@ -221,11 +221,11 @@ every write (no triggers — embeddings are computed in app code):
 
 ## HTTP API
 
-Beyond MCP, the hub exposes a token-authenticated search endpoint. Create a
+Beyond MCP, ClaudexHub exposes a token-authenticated search endpoint. Create a
 token at `/settings/tokens`, then:
 
 ```bash
-curl -H "Authorization: Bearer cxh_…" \
+curl -H "Authorization: Bearer clx_…" \
   "http://localhost:3000/api/v1/search?q=kakao%20cookie&limit=5"
 ```
 
@@ -245,9 +245,9 @@ locally:
 // agent MCP config (HTTP transport)
 {
   "mcpServers": {
-    "context-hub": {
+    "claudexhub": {
       "url": "https://<your-host>/api/mcp",
-      "headers": { "Authorization": "Bearer cxh_…" }
+      "headers": { "Authorization": "Bearer clx_…" }
     }
   }
 }

TODO.md

@@ -83,8 +83,8 @@ Remaining:
 
 ## Phase 10 — packaging & DX
 
-Done: publish-ready package (MIT, metadata, `files`, `context-hub` +
-`context-hub-cli` bins, shebangs, `prepublishOnly`), one-command `init`
+Done: publish-ready package (MIT, metadata, `files`, `claudexhub` +
+`claudexhub-cli` bins, shebangs, `prepublishOnly`), one-command `init`
 (schema + seed), example Claude Code / Cursor MCP configs, and an npx quickstart.
 
 Remaining: