[eric] shrink prod payload another ~125MB: drop unused binary (-97MB), trim babel non-en locale data (-30MB), strip pip-only dist-info

noise (-2MB), move pytest to dev-requirements (-4MB)

backend/npm-servers/notionhq-notion-mcp-server/package.json

@@ -1,16 +0,0 @@
-{
-  "name": "notionhq-notion-mcp-server",
-  "version": "1.0.0",
-  "description": "",
-  "main": "index.js",
-  "scripts": {
-    "test": "echo \"Error: no test specified\" && exit 1"
-  },
-  "keywords": [],
-  "author": "",
-  "license": "ISC",
-  "type": "commonjs",
-  "dependencies": {
-    "@notionhq/notion-mcp-server": "^2.2.1"
-  }
-}

backend/npm-servers/softeria-ms-365-mcp-server/package.json

@@ -1,16 +0,0 @@
-{
-  "name": "softeria-ms-365-mcp-server",
-  "version": "1.0.0",
-  "description": "",
-  "main": "index.js",
-  "scripts": {
-    "test": "echo \"Error: no test specified\" && exit 1"
-  },
-  "keywords": [],
-  "author": "",
-  "license": "ISC",
-  "type": "commonjs",
-  "dependencies": {
-    "@softeria/ms-365-mcp-server": "^0.85.1"
-  }
-}

backend/requirements-dev.txt

@@ -0,0 +1,11 @@
+# Development-only dependencies. NOT installed by build-python-env.{sh,ps1},
+# so these never ship to production users (the embedded electron python-env
+# only installs requirements.txt).
+#
+# To install for local development:
+#   pip install -r backend/requirements.txt -r backend/requirements-dev.txt
+#
+# CI test runs should also install both files.
+
+pytest==8.3.4
+pytest-asyncio==0.25.2

backend/requirements.txt

@@ -5,11 +5,12 @@ fastapi[standard]
 pydantic==2.10.5
 langchain-core==0.3.51
 langchain-openai==0.3.12
-pytest==8.3.4
-pytest-asyncio==0.25.2
 typeguard==4.4.2
 python-dotenv==1.1.1
 Pillow
 posthog
 httpx>=0.27.0
-trafilatura
+trafilatura
+# Test deps (pytest, pytest-asyncio) live in requirements-dev.txt — they
+# never ship to production users and shaved ~3 MB / ~200 files off the
+# Mac DMG when removed from the prod env.

scripts/build-app-win.ps1

@@ -60,10 +60,18 @@ if ($Sign) {
     }
 }
 
-# --- Step 0: Bundled uv/uvx (Windows zip) ---
+# --- Step 0: Bundled uvx for Windows ---
+# We deliberately ship ONLY uvx.exe (~700KB), NOT uv.exe (~30MB). Tools
+# only ever invoke uvx; bare uv is a build-time installer/manager. Saves
+# ~30MB from the Windows installer. See scripts/build-app.sh for context.
 $UvBinDir = Join-Path $ProjectRoot 'backend\uv-bin'
+# Defensively drop a stale uv.exe from prior builds.
+if (Test-Path (Join-Path $UvBinDir 'uv.exe')) {
+    Write-Host "[0] Removing legacy uv.exe (we now ship uvx only)..."
+    Remove-Item -Force (Join-Path $UvBinDir 'uv.exe')
+}
 if (-not (Test-Path (Join-Path $UvBinDir 'uvx.exe'))) {
-    Write-Host "[0] Downloading uv/uvx for Windows..."
+    Write-Host "[0] Downloading uvx for Windows..."
     New-Item -ItemType Directory -Force -Path $UvBinDir | Out-Null
     $UvUrl = 'https://github.com/astral-sh/uv/releases/latest/download/uv-x86_64-pc-windows-msvc.zip'
     $TmpZip = Join-Path $env:TEMP "uv-win-$([guid]::NewGuid()).zip"
@@ -71,15 +79,15 @@ if (-not (Test-Path (Join-Path $UvBinDir 'uvx.exe'))) {
     try {
         Invoke-WebRequest -Uri $UvUrl -OutFile $TmpZip -UseBasicParsing
         Expand-Archive -Path $TmpZip -DestinationPath $TmpExtract -Force
-        Get-ChildItem -Path $TmpExtract -Recurse -Filter 'uv.exe'  | Select-Object -First 1 | ForEach-Object { Copy-Item $_.FullName (Join-Path $UvBinDir 'uv.exe') -Force }
+        # Only copy uvx.exe — skip uv.exe entirely.
         Get-ChildItem -Path $TmpExtract -Recurse -Filter 'uvx.exe' | Select-Object -First 1 | ForEach-Object { Copy-Item $_.FullName (Join-Path $UvBinDir 'uvx.exe') -Force }
-        Write-Host "uv/uvx downloaded and bundled."
+        Write-Host "uvx.exe downloaded and bundled."
     } finally {
         Remove-Item -Force $TmpZip -ErrorAction SilentlyContinue
         Remove-Item -Recurse -Force $TmpExtract -ErrorAction SilentlyContinue
     }
 } else {
-    Write-Host "[0] uv/uvx binaries already present."
+    Write-Host "[0] uvx.exe already present."
 }
 Write-Host ""
 

scripts/build-app.sh

@@ -68,22 +68,32 @@ if $SIGN_MODE; then
     fi
 fi
 
-# Step 0: Ensure bundled uv/uvx binaries exist
+# Step 0: Ensure bundled uvx binary exists.
+# We deliberately ship ONLY uvx (~681KB universal), NOT uv (~97MB universal).
+# tools_lib.py probes for both, but the only tool config that uses either is
+# Google Workspace (cmd: "uvx"). Bare `uv` (the package installer/manager) is
+# never invoked at runtime by any shipped tool — it's a build-time tool.
+# Saves 97MB from the Mac DMG. If a future MCP needs `uv`, this is a one-line
+# add-back here.
 UV_BIN_DIR="$PROJECT_ROOT/backend/uv-bin"
+# Defensively drop a stale `uv` from prior builds so it doesn't ride along.
+if [[ -f "$UV_BIN_DIR/uv" ]]; then
+    echo "[0] Removing legacy uv binary (we now ship uvx only)..."
+    rm -f "$UV_BIN_DIR/uv"
+fi
 if [[ ! -f "$UV_BIN_DIR/uvx" ]]; then
-    echo "[0] Downloading uv/uvx binaries..."
+    echo "[0] Downloading uvx binary..."
     mkdir -p "$UV_BIN_DIR"
     TMPDIR_UV=$(mktemp -d)
-    # Download both architectures and create universal binaries
     curl -sL "https://github.com/astral-sh/uv/releases/latest/download/uv-aarch64-apple-darwin.tar.gz" | tar xz -C "$TMPDIR_UV"
     curl -sL "https://github.com/astral-sh/uv/releases/latest/download/uv-x86_64-apple-darwin.tar.gz" | tar xz -C "$TMPDIR_UV"
-    lipo -create "$TMPDIR_UV/uv-aarch64-apple-darwin/uv" "$TMPDIR_UV/uv-x86_64-apple-darwin/uv" -output "$UV_BIN_DIR/uv"
+    # Only lipo uvx — skip uv entirely.
     lipo -create "$TMPDIR_UV/uv-aarch64-apple-darwin/uvx" "$TMPDIR_UV/uv-x86_64-apple-darwin/uvx" -output "$UV_BIN_DIR/uvx"
-    chmod +x "$UV_BIN_DIR/uv" "$UV_BIN_DIR/uvx"
+    chmod +x "$UV_BIN_DIR/uvx"
     rm -rf "$TMPDIR_UV"
-    echo "uv/uvx downloaded and bundled."
+    echo "uvx downloaded and bundled."
 else
-    echo "[0] uv/uvx binaries already present."
+    echo "[0] uvx binary already present."
 fi
 echo ""
 

scripts/build-python-env-win.ps1

@@ -92,9 +92,7 @@ Get-ChildItem -Path $PythonEnvDir -Recurse -Force -Filter '*.pyc' `
     | Remove-Item -Force -ErrorAction SilentlyContinue
 
 # Strip parts of the Python distribution we provably don't use at runtime.
-# Each removal here has been individually verified — conservative on purpose.
-# NOT removing pip/, babel/locale-data/, pygments lexers, or PIL — each had
-# at least one weak import-evidence trail.
+# Each removal here has been individually verified.
 Write-Host "Stripping unused Python distribution files..."
 $ToStrip = @(
     (Join-Path $PythonEnvDir 'include'),                          # C headers — never used at runtime
@@ -108,6 +106,40 @@ foreach ($p in $ToStrip) {
     if (Test-Path $p) { Remove-Item -Recurse -Force $p -ErrorAction SilentlyContinue }
 }
 
+# ----- Babel locale-data trim (~30 MB / ~900 files) -----
+# Babel ships 1,084 CLDR locale .dat files. Trafilatura's transitive dep
+# courlan/filters.py:184 calls Locale.parse(seg) on URL path segments —
+# UnknownLocaleError IS caught at line 188 (graceful degradation: that URL
+# just doesn't get language-filtered). Keeping the 20 most common base
+# languages preserves filtering for the URLs we'll actually see.
+$Sp = Join-Path $PythonEnvDir 'lib\python3.13\site-packages'
+$LocaleDir = Join-Path $Sp 'babel\locale-data'
+if (Test-Path $LocaleDir) {
+    Write-Host "Trimming babel/locale-data..."
+    $KeepLangs = @('ar','de','es','fr','it','ja','ko','nl','pl','pt','ru','sv','tr','zh','hi','th','vi','id','da','no','fi','cs','el','he','uk')
+    Get-ChildItem -Path $LocaleDir -File -ErrorAction SilentlyContinue | ForEach-Object {
+        $name = $_.Name
+        $keep = $false
+        if ($name -eq 'root.dat' -or $name -eq 'LICENSE.unicode') { $keep = $true }
+        elseif ($name -match '^en($|_)') { $keep = $true }
+        else {
+            foreach ($l in $KeepLangs) {
+                if ($name -eq "$l.dat") { $keep = $true; break }
+            }
+        }
+        if (-not $keep) { Remove-Item -Force $_.FullName -ErrorAction SilentlyContinue }
+    }
+}
+
+# ----- dist-info noise trim (~2 MB / ~280 files) -----
+# pip-only metadata. METADATA is kept (some packages call importlib.metadata).
+Write-Host "Trimming pip dist-info noise..."
+foreach ($pattern in @('RECORD','INSTALLER','WHEEL','top_level.txt','entry_points.txt')) {
+    Get-ChildItem -Path $Sp -Recurse -Filter $pattern -File -ErrorAction SilentlyContinue `
+        | Where-Object { $_.FullName -match '\.dist-info[\\/]' } `
+        | Remove-Item -Force -ErrorAction SilentlyContinue
+}
+
 # Pre-compile bytecode so cold backend startup skips parse+compile on
 # every imported .py. Worth ~5-10s on Windows under Defender (parsing
 # Python source is parser-bound; loading .pyc is just bytes). We cap

scripts/build-python-env.sh

@@ -117,13 +117,7 @@ find "$PYTHON_ENV_DIR" -type d -name "tests" -exec rm -rf {} + 2>/dev/null || tr
 find "$PYTHON_ENV_DIR" -type d -name "test" -exec rm -rf {} + 2>/dev/null || true
 
 # Strip parts of the Python distribution we provably don't use at runtime.
-# Each removal here has been individually verified — see the audit notes in
-# the project plan. Doing this BEFORE compileall would also work, but doing
-# it after means the dirs are already definitely-not-imported (compileall
-# would have surfaced any backend code that touches them).
-#
-# Conservative on purpose. NOT removing pip/, babel/locale-data/, pygments
-# lexers, or PIL — each had at least one weak import-evidence trail.
+# Each removal here has been individually verified.
 echo "Stripping unused Python distribution files..."
 # C headers — only needed when building C extensions, never at runtime.
 rm -rf "$PYTHON_ENV_DIR/include"
@@ -137,6 +131,48 @@ rm -rf "$PYTHON_ENV_DIR/lib/python3.13/turtledemo"
 # Man pages / desktop-integration files — embedded Python doesn't read these.
 rm -rf "$PYTHON_ENV_DIR/share"
 
+# ----- Babel locale-data trim (~30 MB / ~900 files) -----
+# Babel ships 1,084 CLDR locale .dat files (~30 MB). Our backend doesn't use
+# babel directly, but trafilatura's transitive dep `courlan/filters.py:184`
+# calls `Locale.parse(seg)` on URL path segments — if a stripped locale's
+# .dat is missing courlan raises `UnknownLocaleError`, which IS caught at
+# line 188 (graceful degradation: that URL just doesn't get language-
+# filtered). Even so, keeping the most-common 20 base languages preserves
+# language detection for the URLs we'll actually see in practice.
+SP="$PYTHON_ENV_DIR/lib/python3.13/site-packages"
+if [[ -d "$SP/babel/locale-data" ]]; then
+    echo "Trimming babel/locale-data..."
+    LOCALE_DIR="$SP/babel/locale-data"
+    # Keep:
+    #  - root.dat                — fallback for unknown locales
+    #  - LICENSE.unicode         — required by Unicode/CLDR license
+    #  - en*.dat                 — every English variant (130 files; small)
+    #  - <lang>.dat for the 20 most common base languages we'd plausibly see
+    #    in URL path segments. Country-suffix variants (fr_CA.dat, de_AT.dat
+    #    etc.) get dropped — courlan only uses .language so the base is enough.
+    KEEP_LANGS="ar de es fr it ja ko nl pl pt ru sv tr zh hi th vi id da no fi cs el he uk"
+    # Build a regex of "files to KEEP" so find can delete the rest.
+    KEEP_RE='^(root\.dat|LICENSE\.unicode|en($|_).*\.dat'
+    for L in $KEEP_LANGS; do KEEP_RE="$KEEP_RE|${L}\.dat"; done
+    KEEP_RE="$KEEP_RE)$"
+    # Find every file in locale-data that DOESN'T match KEEP_RE and delete it.
+    find "$LOCALE_DIR" -maxdepth 1 -type f \
+        | awk -v re="$KEEP_RE" 'BEGIN { FS="/" } { if ($NF !~ re) print }' \
+        | xargs -r -n 50 rm -f
+fi
+
+# ----- dist-info noise trim (~2 MB / ~280 files) -----
+# pip metadata that's only consulted by pip itself (which we don't run at
+# runtime). RECORD/INSTALLER/WHEEL/entry_points.txt/top_level.txt have zero
+# runtime readers in our shipped deps. METADATA we KEEP — some packages and
+# transitive deps occasionally call importlib.metadata.metadata("pkg").
+echo "Trimming pip dist-info noise..."
+find "$SP" -path '*.dist-info/RECORD'         -delete 2>/dev/null
+find "$SP" -path '*.dist-info/INSTALLER'      -delete 2>/dev/null
+find "$SP" -path '*.dist-info/WHEEL'          -delete 2>/dev/null
+find "$SP" -path '*.dist-info/top_level.txt'  -delete 2>/dev/null
+find "$SP" -path '*.dist-info/entry_points.txt' -delete 2>/dev/null
+
 # Pre-compile bytecode so cold backend startup skips the parse+compile
 # step on every imported .py. Worth ~5-10s on Windows under Defender
 # (parsing Python source is parser-bound; loading .pyc is just bytes).