@@ -24,7 +24,6 @@ ../settings/programs/docker.nix
     ../settings/gaming.nix
 
     ../roles/desktop.nix
-    ../roles/llm-server.nix
     ../roles/nix-remote-builder.nix
     ../roles/home.nix
     ../roles/syncthing.nix

@@ -1,92 +0,0 @@
-{
-  config,
-  lib,
-  pkgs,
-  ...
-}:
-let
-  inherit (config.networking) domain fqdn;
-  ls = config.services.llama-swap;
-in
-{
-  environment.systemPackages = [
-    pkgs.python3Packages.huggingface-hub
-  ];
-
-  services.llama-swap = {
-    enable = true;
-    listenAddress = "[::]";
-    port = 11343;
-    settings =
-      let
-        llama-cpp = pkgs.llama-cpp.override { cudaSupport = true; };
-        llama-server = lib.getExe' llama-cpp "llama-server";
-        llama = "${llama-server} --port \${PORT}";
-        next_edit = ''
-          --spec-type ngram-mod
-          --spec-ngram-mod-n-match 24
-          --spec-ngram-mod-n-min 24
-          --spec-ngram-mod-n-max 32
-          --no-webui
-        '';
-      in
-      {
-        healthCheckTimeout = 60;
-        models = {
-          sweep-next-edit = {
-            name = "Sweep Next Edit 1.5B";
-            cmd = "${llama} ${next_edit} --hf-repo sweepai/sweep-next-edit-1.5B";
-            aliases = [ "sweep-next-edit-1.5B" ];
-          };
-          sweep-next-edit-v2 = {
-            name = "Sweep Next Edit V2 7B";
-            cmd = "${llama} ${next_edit} --hf-repo henrik3/sweep-next-edit-v2-7B-GGUF";
-            aliases = [ "sweep-next-edit-v2-7B" ];
-          };
-          zeta20 = {
-            name = "Zeta 2.0";
-            cmd = "${llama} ${next_edit} --hf-repo bartowski/zed-industries_zeta-2-GGUF";
-            aliases = [ "zeta-2" ];
-          };
-          zeta21 = {
-            name = "Zeta 2.1";
-            cmd = "${llama} ${next_edit} --hf-repo mradermacher/zeta-2.1-GGUF";
-            aliases = [ "zeta-2.1" ];
-          };
-        };
-        groups = {
-          edit_prediction = {
-            swap = true;
-            members = [
-              "sweep-next-edit"
-              "sweep-next-edit-v2"
-              "zeta20"
-              "zeta21"
-            ];
-          };
-        };
-        includeAliasesInList = true;
-        hooks.on_startup.preload = [ "sweep-next-edit" ];
-      };
-  };
-  systemd.services.llama-swap = {
-    environment.XDG_CACHE_HOME = "/var/cache/llama-swap";
-    serviceConfig.CacheDirectory = "llama-swap";
-  };
-
-  services.caddy = {
-    enable = true;
-
-    globalConfig = ''
-      acme_ca https://acme.${domain}/directory
-    '';
-
-    virtualHosts = {
-      "llama.${fqdn}" = {
-        extraConfig = ''
-          reverse_proxy ${ls.listenAddress}:${toString ls.port}
-        '';
-      };
-    };
-  };
-}

@@ -3,9 +3,6 @@ config,
   pkgs,
   ...
 }:
-let
-  nixfiles = config.home.homeDirectory + "/projects/alin.ovh/nixfiles";
-in
 {
   imports = [
     ../settings/base.nix
@@ -23,16 +20,13 @@ ../settings/darwin.nix
     ../settings/emacs.nix
     ../settings/fish.nix
     ../settings/git.nix
-    ../settings/local-llm.nix
     ../settings/nixpkgs.nix
-    ../settings/mcp.nix
     ../settings/ssh.nix
     ../settings/workstation.nix
     ../settings/user-interface.nix
     ../settings/nix.nix
     ../settings/zed.nix
     ../../private/ssh.nix
-    ../../private/llm.nix
   ];
 
   home.username = "alin";
@@ -134,10 +128,5 @@ };
     "com.apple.timemachine" = {
       DoNotOfferNewDisksForBackup = true;
     };
-  };
-
-  services.llama-swap = {
-    configFile = "${nixfiles}/user/settings/llama/swap-config.yml";
-    extraArgs = [ "--watch-config" ];
   };
 }

@@ -1,165 +0,0 @@
-{
-  config,
-  lib,
-  pkgs,
-  ...
-}:
-let
-  cfg = config.services.llama-swap;
-  settingsFormat = pkgs.formats.yaml { };
-  configFile =
-    if cfg.configFile != null then
-      cfg.configFile
-    else
-      settingsFormat.generate "config.yaml" cfg.settings;
-
-  args = [
-    "--listen=${cfg.listenAddress}:${toString cfg.port}"
-    "--config=${configFile}"
-  ]
-  ++ cfg.extraArgs;
-in
-{
-  options.services.llama-swap = {
-    enable = lib.mkEnableOption "the llama-swap service";
-
-    package = lib.mkPackageOption pkgs "llama-swap" { };
-
-    packages = lib.mkOption {
-      default = lib.optionals (cfg.configFile != null) [ pkgs.llama-cpp ];
-      defaultText = lib.literalExpression ''
-        lib.optionals (cfg.configFile != null) [ pkgs.llama-cpp ]
-      '';
-      example = lib.literalExpression ''
-        [
-          pkgs.llama-cpp-rocm
-          pkgs.vllm
-        ]
-      '';
-      type = lib.types.listOf lib.types.package;
-      description = ''
-        Packages to add to llama-swap process's PATH.
-        Useful if using `configFile` instead of `settings`.
-      '';
-    };
-
-    exposePackages = lib.mkOption {
-      default = cfg.configFile != null;
-      defaultText = lib.literalExpression "cfg.configFile != null";
-      type = lib.types.bool;
-      description = ''
-        Whether to additionally add `packages` to user profile.
-        Useful when using `configFile` instead of `settings`.
-      '';
-    };
-
-    listenAddress = lib.mkOption {
-      type = lib.types.str;
-      default = "localhost";
-      example = "0.0.0.0";
-      description = ''
-        Address that llama-swap listens on.
-      '';
-    };
-
-    port = lib.mkOption {
-      default = 8080;
-      example = 11343;
-      type = lib.types.port;
-      description = ''
-        Port that llama-swap listens on.
-      '';
-    };
-
-    configFile = lib.mkOption {
-      type = with lib.types; nullOr (either path str);
-      default = null;
-      description = ''
-        Optional configuration file path to use instead of
-        generating from settings
-      '';
-    };
-
-    extraArgs = lib.mkOption {
-      type = with lib.types; listOf str;
-      default = [ ];
-      description = ''
-        Additional arguments to pass on llama-swap startup
-      '';
-    };
-
-    settings = lib.mkOption {
-      type = lib.types.submodule { freeformType = settingsFormat.type; };
-      description = ''
-        llama-swap configuration. Refer to the [llama-swap example configuration](https://github.com/mostlygeek/llama-swap/blob/main/config.example.yaml)
-        for details on supported values.
-      '';
-      example = lib.literalExpression ''
-        let
-          llama-cpp = pkgs.llama-cpp.override { rocmSupport = true; };
-          llama-server = lib.getExe' llama-cpp "llama-server";
-        in
-        {
-          healthCheckTimeout = 60;
-          models = {
-            "some-model" = {
-              cmd = "''${llama-server} --port ''${PORT} -m /var/lib/llama-cpp/models/some-model.gguf -ngl 0 --no-webui";
-              aliases = [
-                "the-best"
-              ];
-            };
-            "other-model" = {
-              proxy = "http://127.0.0.1:5555";
-              cmd = "''${llama-server} --port 5555 -m /var/lib/llama-cpp/models/other-model.gguf -ngl 0 -c 4096 -np 4 --no-webui";
-              concurrencyLimit = 4;
-            };
-          };
-        };
-      '';
-    };
-  };
-  config = lib.mkIf cfg.enable {
-    assertions = [
-      {
-        assertion = cfg.configFile != null -> (builtins.attrNames cfg.settings) == [ ];
-        message = "configFile is incompatible with non-empty settings";
-      }
-    ];
-
-    # untested
-    systemd.user.services.llama-swap = {
-      description = "Model swapping for LLaMA C++ Server (or any local OpenAPI compatible server)";
-
-      serviceConfig = {
-        Type = "exec";
-        ExecStart = "${lib.getExe cfg.package} ${lib.escapeShellArgs args}";
-        Restart = "on-failure";
-        RestartSec = 3;
-
-        # for GPU acceleration
-        PrivateDevices = false;
-      };
-      path = cfg.packages;
-    };
-
-    launchd.agents.llama-swap = {
-      enable = true;
-      config = {
-        ProgramArguments = [ (lib.getExe cfg.package) ] ++ args;
-        KeepAlive = {
-          Crashed = true;
-          SuccessfulExit = false;
-        };
-        ProcessType = "Background";
-        EnvironmentVariables = {
-          PATH = lib.makeBinPath cfg.packages;
-        };
-      };
-    };
-
-    home.packages = [
-      cfg.package
-    ]
-    ++ (lib.optionals cfg.exposePackages cfg.packages);
-  };
-}

@@ -23,10 +23,6 @@ hugo.localhost {
 	reverse_proxy http://localhost:1313
 }
 
-llama.localhost, http://llama.localhost {
-	reverse_proxy http://localhost:11343
-}
-
 searchix.localhost {
 	handle_errors {
 		rewrite * /error.html

@@ -1,75 +0,0 @@
-macros:
-  llama: |
-    llama-server --port ${PORT}
-  mlx: |
-    mlx_lm.server --port ${PORT}
-
-  # - small `n` are not recommended
-  # - MoEs require long drafts
-  # - dense models: can reduce `--draft-min` and `--draft-max`
-  next_edit: |
-    --spec-type ngram-mod
-    --spec-ngram-mod-n-match 24
-    --spec-ngram-mod-n-min 24
-    --spec-ngram-mod-n-max 32
-
-models:
-  sweep-next-edit-1.5b:
-    name: Sweep Next-Edit 1.5B
-    unlisted: true
-    aliases:
-      - sweep-next-edit
-    cmd: |
-      ${llama} --hf-repo sweepai/sweep-next-edit-1.5B
-      ${next_edit}
-  sweep-next-edit-v2-7b:
-    name: Sweep Next-Edit v2 7B
-    unlisted: true
-    aliases:
-      - sweep-next-edit-v2
-    cmd: |
-      ${llama} --hf-repo henrik3/sweep-next-edit-v2-7B-GGUF
-      ${next_edit}
-  gpt-oss-20b:
-    name: GPT OSS (20b)
-    cmd: |
-      ${llama} --hf-repo unsloth/gpt-oss-20b-GGUF
-  gpt-oss-120b:
-    name: GPT OSS (120b)
-    cmd: |
-      ${llama} --hf-repo unsloth/gpt-oss-120b-GGUF
-  qwen3.6-27b:
-    name: Qwen 3.6 27B
-    cmd: |
-      ${llama} --hf-repo unsloth/Qwen3.6-27B-MTP-GGUF:Q8_0
-      --spec-type draft-mtp
-      --flash-attn on
-      --ctx-size 65536
-  gemma-4-coder-fable5-composer2.5:
-    name: Gemma 4 Coder (Fable 5, Composer 2.5)
-    useModelName: yuxinlu1/gemma-4-12B-coder-fable5-composer2.5-v1-GGUF
-    cmd: |
-      ${llama} --hf-repo yuxinlu1/gemma-4-12B-coder-fable5-composer2.5-v1-GGUF:Q8_0
-
-groups:
-  edit_prediction:
-    persistent: true # not unloaded by other groups
-    swap: true # only run one model from this group
-    exclusive: false
-    members:
-      - sweep-next-edit-1.5b
-      - sweep-next-edit-v2-7b
-
-hooks:
-  on_startup:
-    preload:
-      - sweep-next-edit-v2-7b
-
-peers:
-  prefect:
-    proxy: http://prefect:11343
-    models:
-      - sweep-next-edit
-      - sweep-next-edit-v2
-      - zeta20
-      - zeta21

@@ -1,23 +0,0 @@
-{
-  pkgs,
-  ...
-}:
-{
-  imports = [
-    ../modules/llama-swap.nix
-  ];
-
-  home.packages = [
-    pkgs.python3Packages.huggingface-hub
-  ];
-
-  services.llama-swap = {
-    enable = true;
-    packages = with pkgs; [
-      llama-cpp
-      python3Packages.mlx-lm
-    ];
-    listenAddress = "[::1]";
-    port = 11343;
-  };
-}

@@ -1,31 +0,0 @@
-{
-  lib,
-  pkgs,
-  ...
-}:
-{
-  # servers with secrets should go in ../../private/llm.nix
-  programs.mcp = {
-    enable = true;
-    servers = {
-      nixos = {
-        command = "nix";
-        args = [
-          "run"
-          "github:utensils/mcp-nixos"
-          "--"
-        ];
-      };
-    }
-    // (lib.optionalAttrs pkgs.stdenv.isDarwin {
-      dash = {
-        command = "${pkgs.uv}/bin/uvx";
-        args = [
-          "--from"
-          "git+https://github.com/Kapeli/dash-mcp-server.git"
-          "dash-mcp-server"
-        ];
-      };
-    });
-  };
-}

@@ -6,7 +6,6 @@ }:
 {
   programs.zed-editor = {
     enable = true;
-    enableMcpIntegration = true;
     userSettings = {
       base_keymap = "Emacs";
       vim_mode = true;