levelcodeai
diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
@@ -30,7 +30,7 @@ jobs:
           node-version: "24"
       - name: Extension unit tests
         run: |
-          cd extensions/atom-ai
+          cd extensions/levelcode-ai
           for t in test/*.test.js; do node "$t"; done
 
   build:

diff --git a/CLAUDE.md b/CLAUDE.md
@@ -15,7 +15,7 @@ lives in the tracked repo and is re-applied onto a clean clone:
 
 | Ours (edit here, tracked) | Gets installed into (generated) | By |
 | --- | --- | --- |
-| `extensions/atom-npp-pack/`, `extensions/atom-ai/` | `vscode/extensions/*` | `apply-branding.mjs` (copy) |
+| `extensions/levelcode-npp-pack/`, `extensions/levelcode-ai/` | `vscode/extensions/*` | `apply-branding.mjs` (copy) |
 | `branding/product.overlay.json`, `branding/icons/` | `vscode/product.json`, `vscode/resources/...` | `apply-branding.mjs` (merge/copy) |
 | `patches/levelcode-core.patch` | edits to `vscode/src/**`, `vscode/build/**` | `bootstrap.sh` (`git apply`) |
 
@@ -33,12 +33,12 @@ docs/CORE-PATCHES.md        log of every edit inside vscode/ (tagged // [LevelCo
 docs/M0-RUNBOOK.md, EXIT-TEST.md, M1-SPEC.md
 branding/product.overlay.json   LevelCode identity + Open VSX gallery (deep-merged onto product.json)
 branding/icons/             app icon source PNG, generated .icns, 1024 png
-extensions/atom-npp-pack/   Notepad++ power-editing pack (plain JS, no build step)
-extensions/atom-ai/         native Claude AI (chat, inline completion, providers, edit-with-diff, LM provider)
-extensions/atom-themes/     signature One Dark / One Light themes (JSON, default via configurationDefaults)
-extensions/atom-hackability/ user init script + Atom/NPP keymap presets + package generator & hot-reload dev loader
-extensions/atom-sync/       'levelcode' auth provider that lights up the built-in Settings Sync (LevelCode Sync, S0)
-extensions/atom-updater/    notify-only update checker (polls the update feed; never auto-applies)
+extensions/levelcode-npp-pack/   Notepad++ power-editing pack (plain JS, no build step)
+extensions/levelcode-ai/         native Claude AI (chat, inline completion, providers, edit-with-diff, LM provider)
+extensions/levelcode-themes/     signature One Dark / One Light themes (JSON, default via configurationDefaults)
+extensions/levelcode-hackability/ user init script + Atom/NPP keymap presets + package generator & hot-reload dev loader
+extensions/levelcode-sync/       'levelcode' auth provider that lights up the built-in Settings Sync (LevelCode Sync, S0)
+extensions/levelcode-updater/    notify-only update checker (polls the update feed; never auto-applies)
 patches/levelcode-core.patch     our core source edits, applied on bootstrap
 scripts/                    bootstrap.sh, apply-branding.mjs, run-dev.sh, build-macos.sh, make-dmg.sh, make-icon.sh; atom (CLI launcher) + install-level.sh
 tools/                      dependency-free reference servers: sync-server (/v1 Settings-Sync), update-server (/api/update feed)
@@ -83,17 +83,17 @@ Tracked in `patches/levelcode-core.patch`, tagged with `// [LevelCode]`. Find th
 ## Settings shipped as defaults (via extension `configurationDefaults` or core patch)
 
 - `files.hotExit = onExitAndWindowClose` (core patch — application-scoped).
-- `workbench.editorLargeFileConfirmation = 2048`, `chat.commandCenter.enabled = false`, `chat.disableAIFeatures = true` (atom-npp-pack).
+- `workbench.editorLargeFileConfirmation = 2048`, `chat.commandCenter.enabled = false`, `chat.disableAIFeatures = true` (levelcode-npp-pack).
 - Extensions can only override **machine-overridable / window / resource / language-overridable** scoped settings — NOT application/machine. (That's why hotExit needed a core patch.)
 
 ## Feature status
 
-**M1 — Notepad++ pack (`extensions/atom-npp-pack/`, all done, plain JS):**
+**M1 — Notepad++ pack (`extensions/levelcode-npp-pack/`, all done, plain JS):**
 macros (`Cmd+Shift+R`/`Cmd+Alt+P`), Duplicate file (`Cmd+D` / Explorer menu), Sublime hot-exit,
 line operations (sort/dedup/case/…), column incrementing numbers, encoding/EOL status-bar toggle,
 big-file mode badge. Files: extension.js + fileOps/lineOps/columnOps/encodingEol/bigFile.js.
 
-**M2 — native AI (`extensions/atom-ai/`, working):**
+**M2 — native AI (`extensions/levelcode-ai/`, working):**
 - `providers/` — **multi-provider BYOK (P1)**. A registry data table (`providers/index.js` `PROVIDERS` +
   `getProvider`/`normId`/`secretStorageKey`/`isInsecureCustomUrl`) dispatches `streamChat()`/`complete()` on the
   row's `kind`: **Anthropic** keeps its native adapter (`providers/anthropic.js` — prompt caching + full tool-use,
@@ -147,13 +147,13 @@ big-file mode badge. Files: extension.js + fileOps/lineOps/columnOps/encodingEol
   multi-session core effort. We use the diff-tab review instead.
 - Built since M2 shipped: inline tab-completion (`inlineComplete.js`), multi-file + auto-retrieval chat context
   (`gatherAutoContext`), chat moved to the right side bar, typewriter streaming, `make-dmg.sh` packaging.
-- M3 so far: One Dark/Light themes (`atom-themes`); user init script + Atom/Notepad++ keymap presets +
-  package generator with live hot-reload (`atom-hackability`). The Atom/NPP keymap also clears the old M1 leftover.
+- M3 so far: One Dark/Light themes (`levelcode-themes`); user init script + Atom/Notepad++ keymap presets +
+  package generator with live hot-reload (`levelcode-hackability`). The Atom/NPP keymap also clears the old M1 leftover.
 - M4 agentic multi-file tasks: DONE. `agent.js` has the full tool loop — list_files/read_file/search,
   update_plan, edit_file/write_file/**delete_file** (apply-then-review with Keep/Undo + per-turn checkpoint
   restore via `reviewSession.js`), run_command(+background)/read_command_output, ask_user, use_skill — plus the
   M5 auto-verify loop. `delete_file` also enables rename/move (write new path → delete old).
-- LevelCode Sync (S0) + notify-only updater (U0): see `extensions/atom-sync` / `extensions/atom-updater` + `tools/`.
+- LevelCode Sync (S0) + notify-only updater (U0): see `extensions/levelcode-sync` / `extensions/levelcode-updater` + `tools/`.
 - Not yet built: remaining M3 hackability — settings UI, theme studio, VS Code settings import.
 
 ## Conventions

diff --git a/blog/atom-style-package-hot-reload.md b/blog/atom-style-package-hot-reload.md
@@ -169,5 +169,5 @@ Write, save, see it. Welcome back.
 ---
 
 *LevelCode is an AI-native, hackable, Notepad++-powered editor for macOS, built on the MIT
-Code-OSS core. The package generator and hot-reload loader live in the `atom-hackability`
+Code-OSS core. The package generator and hot-reload loader live in the `levelcode-hackability`
 extension, alongside the user init script and the Atom/Notepad++ keymap presets.*
diff --git a/branding/product.overlay.json b/branding/product.overlay.json
@@ -23,15 +23,15 @@
   "urlProtocol": "levelcode",
   "quality": "stable",
 
-  "_syncComment": "LevelCode Sync (built-in Settings Sync + the `levelcode` auth provider from extensions/atom-sync) was REMOVED for now — the managed sync host isn't built and it duplicated the LevelCode Cloud sign-in. Removed at source: the atom-sync extension, this configurationSync.store block, and the Welcome 'Sync' step. Re-add when the managed sync host ships.",
+  "_syncComment": "LevelCode Sync (built-in Settings Sync + the `levelcode` auth provider from extensions/levelcode-sync) was REMOVED for now — the managed sync host isn't built and it duplicated the LevelCode Cloud sign-in. Removed at source: the levelcode-sync extension, this configurationSync.store block, and the Welcome 'Sync' step. Re-add when the managed sync host ships.",
 
   "licenseName": "MIT",
-  "licenseUrl": "https://github.com/levelcode/levelcode/blob/main/LICENSE",
-  "reportIssueUrl": "https://github.com/levelcode/levelcode/issues/new",
+  "licenseUrl": "https://github.com/levelcodeai/levelcode/blob/main/LICENSE",
+  "reportIssueUrl": "https://github.com/levelcodeai/levelcode/issues/new",
 
   "_downloadComment": "Download + update funnel through levelcode.ai (see site/app/download/[arch]/route.ts). downloadUrl is where the notify-only updater's Download button sends the user; releaseNotesUrl backs its 'Release Notes' button. updateUrl is the feed base — the updater GETs {updateUrl}/api/update/{target}/{quality}/{commit}; until that endpoint is live it returns 404 and the updater silently treats it as 'up to date' (notify-only, non-fatal). RELEASE NOTE: levelcode.ai must resolve to the deployed site before these links work.",
   "downloadUrl": "https://levelcode.ai/download",
-  "releaseNotesUrl": "https://github.com/levelcode/levelcode/releases/latest",
+  "releaseNotesUrl": "https://github.com/levelcodeai/levelcode/releases/latest",
   "updateUrl": "https://levelcode.ai",
 
   "enableTelemetry": false,

diff --git a/docs/CORE-PATCHES.md b/docs/CORE-PATCHES.md
@@ -10,7 +10,7 @@ comment in the code, so a rebase onto a newer Code-OSS tag is easy to re-apply/v
 that is *ours* lives in the tracked repo and is re-installed onto a clean clone:
 
 - **Branding** → `branding/product.overlay.json` (+ `branding/icons/`), applied by `apply-branding.mjs`.
-- **Extensions** → `extensions/atom-npp-pack/`, `extensions/atom-ai/` are the **canonical source**;
+- **Extensions** → `extensions/levelcode-npp-pack/`, `extensions/levelcode-ai/` are the **canonical source**;
   `apply-branding.mjs` copies them into `vscode/extensions/`. Edit them *here*, not in `vscode/`.
 - **Core source edits** → captured as `patches/levelcode-core.patch`; `bootstrap.sh` applies it
   (`git apply --3way`) onto the fresh checkout.

diff --git a/docs/M1-SPEC.md b/docs/M1-SPEC.md
@@ -4,11 +4,11 @@ Goal of M1: make LevelCode feel like Notepad++ for the editing muscle-memory Ser
 
 ## Design principle: isolate, don't scatter
 
-Everything that **can** live in a first-party bundled extension **does** — shipped enabled by default as the "Notepad++ Pack" (`extensions/atom-npp-pack/`). This keeps our code out of upstream core files so rebasing onto new Code-OSS stays cheap (PLAN.md §8). We only touch core where the extension API genuinely can't reach: large-file rendering and the encoding/EOL status-bar affordances. Those core touches are kept in clearly marked, self-contained modules.
+Everything that **can** live in a first-party bundled extension **does** — shipped enabled by default as the "Notepad++ Pack" (`extensions/levelcode-npp-pack/`). This keeps our code out of upstream core files so rebasing onto new Code-OSS stays cheap (PLAN.md §8). We only touch core where the extension API genuinely can't reach: large-file rendering and the encoding/EOL status-bar affordances. Those core touches are kept in clearly marked, self-contained modules.
 
 Two buckets:
 
-- **Bucket A — bundled extension** (`atom-npp-pack`): macros, line operations, column-mode extras, NPP keymap. No core edits.
+- **Bucket A — bundled extension** (`levelcode-npp-pack`): macros, line operations, column-mode extras, NPP keymap. No core edits.
 - **Bucket B — core touches** (small, isolated): big-file mode, always-visible encoding/EOL status bar items. Each guarded behind an `_atomPlusPlus` product.json flag so it's easy to diff and toggle.
 
 ---
@@ -73,7 +73,7 @@ Ship a selectable **"Notepad++"** keymap (alongside an "Atom" keymap from M3) so
 
 ## Build order & milestone gates
 
-1. **Scaffold `atom-npp-pack`** bundled extension + "Notepad++ Pack" settings namespace (`levelcode.*`). Wire it into the build's built-in extensions list.
+1. **Scaffold `levelcode-npp-pack`** bundled extension + "Notepad++ Pack" settings namespace (`levelcode.*`). Wire it into the build's built-in extensions list.
 2. **Macros** (headline) → demo + exit test.
 3. **Line ops + column extras** (fast wins, pure extension) → exit tests.
 4. **Encoding/EOL status bar** (Bucket B, small core touch) → exit test.

diff --git a/docs/RELEASING.md b/docs/RELEASING.md
@@ -88,7 +88,7 @@ from **levelcode.ai/download**.
 
 ## 5. Point the update feed at the release
 
-The notify-only updater (`extensions/atom-updater`) polls a feed from `tools/update-server`. After
+The notify-only updater (`extensions/levelcode-updater`) polls a feed from `tools/update-server`. After
 publishing, update the feed's release entry (version + the release-asset URLs above) so running installs
 see the new version. Keep the feed version in lockstep with the tag.
 

diff --git a/docs/levelcode-agent-sketch-design.md b/docs/levelcode-agent-sketch-design.md
@@ -22,7 +22,7 @@ Everything below sits on the multi-provider layer already shipped:
 ## 1. The SK1 skeleton (shipped)
 
 ```
-extensions/atom-ai/
+extensions/levelcode-ai/
   sketch.js                  controller: panel, persistence, THE RUNNER (reentrancy-guarded, abortable)
   sketch/agentCatalog.js     96 agent archetypes in 13 palette groups (derived from ruflo, MIT © ruvnet — names + one-line roles only)
   sketch/templates.js        ready-made topologies — incl. the "Design a Key-Value Store" hero template   [unit-tested]

diff --git a/docs/levelcode-multiprovider-design.md b/docs/levelcode-multiprovider-design.md
@@ -39,12 +39,12 @@ Everything reduces to **OpenAI Chat Completions**, **Anthropic Messages** (block
 
 ## 2. Recommended architecture for LevelCode
 
-### 2.1 Module layout — `extensions/atom-ai/providers/`
+### 2.1 Module layout — `extensions/levelcode-ai/providers/`
 
 Today everything is one flat `providers.js` with eight provider-named exports and hard-coded Anthropic assumptions in 4 places. Evolve it into a directory (keep `providers.js` as a thin re-export shim during migration so nothing breaks mid-refactor):
 
 ```
-extensions/atom-ai/providers/
+extensions/levelcode-ai/providers/
   index.js         # registry (PROVIDERS data table) + getProvider(id) + resolveProvider(id) + shared readLines()
   anthropic.js     # native Anthropic Messages adapter (streamClaude/completeClaude/streamClaudeAgentTurn moved here)
   openaiCompat.js  # THE universal adapter — /v1/chat/completions, param'd by {baseURL, apiKey, headers}