systemd: the service manager as a module — holds node-service-manager and answers for the units in both scopes

The holder of the seat the controller seeds under novox/hq ADR 0177. Eight
verbs under the seat's name — units, status, start, stop, restart, enable,
disable, journal — each taking an optional scope, "system" by default or
"user" for the operator account's own manager, reached as
`systemctl --user --machine=<account>@` when the runtime is not that account.
One tool of its own, systemd_failed, for every failed unit in both scopes.
A package, a claim and a bundle; no container, no process: served by the node
tools runtime (ADR 0175) once it exists. `module check` passes against a
controller that carries the seat; the tools type-check against the SDK.

modules/lab/Dockerfile

@@ -10,12 +10,12 @@ ARG RUNTIME_BASE
 FROM ${BUILD_BASE} AS build
 WORKDIR /app/modules/lab
 COPY . .
-RUN node /app/node_modules/typescript/bin/tsc tools/index.ts tools/runs.ts \
+RUN node /app/node_modules/typescript/bin/tsc tools/index.ts tools/runs.ts --rootDir . \
       --module NodeNext --moduleResolution NodeNext --target ES2022 --outDir dist
 
 FROM ${RUNTIME_BASE}
 RUN apt-get update \
- && apt-get install -y --no-install-recommends git make ca-certificates curl \
+ && apt-get install -y --no-install-recommends git make ca-certificates curl python3 file iproute2 sudo \
  && rm -rf /var/lib/apt/lists/*
 RUN curl -fsSL -o /tmp/go.tgz https://go.dev/dl/go1.26.8.linux-amd64.tar.gz \
  && echo "d0f743b33e8d8945e6b1f432edd15785c70507121d6e2a723b21285eddf8b57b  /tmp/go.tgz" | sha256sum -c - \

modules/lab/module.json

@@ -2,7 +2,8 @@
   "module": "lab",
   "version": "1",
   "capabilities": [
-    "container-runtime"
+    "container-runtime",
+    "virtualisation"
   ],
   "own-secrets": {
     "broker": "${dir:mesh-state}/broker"

modules/lab/tools/runs.ts

@@ -100,6 +100,10 @@ node -e ${shellQuote(
     `const fs=require("fs");const f=${JSON.stringify(join(dir, "status.json"))};const s=JSON.parse(fs.readFileSync(f,"utf8"));s.commits=Object.fromEntries(fs.readFileSync(${JSON.stringify(join(dir, "commits.txt"))},"utf8").trim().split("\\n").map(l=>l.split(" ")));fs.writeFileSync(f,JSON.stringify(s,null,2))`,
   )}
 ${setState("building")}
+# The @novox scope resolves from the mesh's own package registry on the forge, as the build machine
+# resolves it; nothing else is asked of it.
+printf '%s\n' ${shellQuote(`@novox:registry=${forge}/api/packages/novox/npm/`)} > ${shellQuote(join(dir, ".npmrc"))}
+export NPM_CONFIG_USERCONFIG=${shellQuote(join(dir, ".npmrc"))}
 for repo in mesh-sdk mesh-tools mesh-lab; do (cd ${shellQuote(dir)}/$repo && npm ci --no-audit --no-fund --loglevel=error); done
 (cd ${shellQuote(dir)}/mesh-sdk && npm run build --if-present)
 (cd ${shellQuote(dir)}/mesh-tools && npm run build --if-present)

modules/systemd/client.ts

@@ -0,0 +1,93 @@
+// systemctl and journalctl, asked in one scope or the other (novox/hq ADR 0177).
+//
+// The system manager is the machine's. The user manager is the operator account's own: reached as
+// `systemctl --user --machine=<account>@` when this process is not that account (the node tools
+// runtime runs as the node's account, root when the host started it), and as plain `--user` when
+// it is. It answers only while the account's manager runs — a login, or lingering enabled.
+
+import { execFile } from "node:child_process";
+import { userInfo } from "node:os";
+
+export type Scope = "system" | "user";
+
+export interface Unit {
+  unit: string;
+  load: string;
+  active: string;
+  sub: string;
+  description: string;
+}
+
+function run(cmd: string, args: string[]): Promise<{ stdout: string; stderr: string; status: number }> {
+  return new Promise((resolve) => {
+    execFile(cmd, args, { maxBuffer: 8 * 1024 * 1024 }, (err, stdout, stderr) => {
+      const status = err && typeof (err as { code?: unknown }).code === "number" ? ((err as { code: number }).code) : err ? 1 : 0;
+      resolve({ stdout: String(stdout ?? ""), stderr: String(stderr ?? "") + (err && !(err as { code?: unknown }).code ? err.message : ""), status });
+    });
+  });
+}
+
+export class ServiceManager {
+  constructor(private readonly account: string) {}
+
+  static fromEnv(env: NodeJS.ProcessEnv): ServiceManager {
+    return new ServiceManager(env.MESH_OPERATOR_ACCOUNT?.trim() || userInfo().username);
+  }
+
+  /** The leading arguments that pick a manager. */
+  scopeArgs(scope: Scope): string[] {
+    if (scope !== "user") return [];
+    return userInfo().username === this.account ? ["--user"] : ["--user", `--machine=${this.account}@`];
+  }
+
+  async systemctl(scope: Scope, ...args: string[]): Promise<{ stdout: string; stderr: string; status: number }> {
+    return run("systemctl", [...this.scopeArgs(scope), ...args]);
+  }
+
+  async units(scope: Scope, pattern?: string): Promise<Unit[]> {
+    const args = ["list-units", "--all", "--no-legend", "--plain", "--no-pager"];
+    if (pattern) args.push(pattern);
+    const { stdout } = await this.systemctl(scope, ...args);
+    return stdout
+      .split("\n")
+      .map((l) => l.trim())
+      .filter(Boolean)
+      .map((l) => {
+        const [unit, load, active, sub, ...rest] = l.split(/\s+/);
+        return { unit, load, active, sub, description: rest.join(" ") };
+      });
+  }
+
+  async status(scope: Scope, unit: string): Promise<Record<string, string>> {
+    const props = ["LoadState", "ActiveState", "SubState", "UnitFileState", "MainPID", "ExecMainStatus", "Description", "FragmentPath"];
+    const { stdout } = await this.systemctl(scope, "show", unit, ...props.map((p) => `--property=${p}`));
+    const out: Record<string, string> = { unit, scope };
+    for (const line of stdout.split("\n")) {
+      const i = line.indexOf("=");
+      if (i > 0) out[line.slice(0, i)] = line.slice(i + 1);
+    }
+    return out;
+  }
+
+  async act(scope: Scope, verb: "start" | "stop" | "restart" | "enable" | "disable", unit: string): Promise<Record<string, unknown>> {
+    const { stderr, status } = await this.systemctl(scope, verb, unit);
+    const after = await this.status(scope, unit);
+    return { unit, scope, verb, ok: status === 0, stderr: stderr.trim(), active: after.ActiveState, boot: after.UnitFileState,
+      note: "a unit the mesh declares is restored to its declared state at the host's next apply" };
+  }
+
+  async journal(scope: Scope, unit: string, lines: number): Promise<{ unit: string; scope: Scope; lines: string[] }> {
+    const args = ["--no-pager", "-n", String(lines), "-u", unit, "-o", "short-iso"];
+    if (scope === "user") {
+      args.unshift(userInfo().username === this.account ? "--user" : `--machine=${this.account}@`, ...(userInfo().username === this.account ? [] : ["--user"]));
+    }
+    const { stdout } = await run("journalctl", args);
+    return { unit, scope, lines: stdout.split("\n").filter(Boolean) };
+  }
+
+  async failed(): Promise<{ system: Unit[]; user: Unit[] }> {
+    const system = (await this.units("system")).filter((u) => u.active === "failed");
+    const user = (await this.units("user").catch(() => [] as Unit[])).filter((u) => u.active === "failed");
+    return { system, user };
+  }
+}

modules/systemd/module.json

@@ -0,0 +1,46 @@
+{
+  "module": "systemd",
+  "version": "1",
+  "capabilities": [
+    "service-manager",
+    "package-manager"
+  ],
+  "claims": [
+    {
+      "name": "node-service-manager",
+      "scope": "node",
+      "serves": [
+        "units",
+        "status",
+        "start",
+        "stop",
+        "restart",
+        "enable",
+        "disable",
+        "journal"
+      ]
+    }
+  ],
+  "tools": [
+    "systemd_failed"
+  ],
+  "resources": [
+    {
+      "id": "package",
+      "type": "package",
+      "package": "systemd"
+    }
+  ],
+  "build": {
+    "artifacts": [
+      {
+        "name": "tools",
+        "kind": "bundle",
+        "language": "typescript",
+        "entrypoints": [
+          "tools/index.js"
+        ]
+      }
+    ]
+  }
+}

modules/systemd/package.json

@@ -0,0 +1,14 @@
+{
+  "name": "@novox/module-systemd",
+  "version": "0.1.0",
+  "description": "systemd \u2014 the machine's service manager as a module: holds node-service-manager and answers for the units in both scopes (novox/hq ADR 0177). The host applies units; this answers about them.",
+  "type": "module",
+  "private": true,
+  "dependencies": {
+    "@novox/mesh-sdk": "^0.1.0"
+  },
+  "devDependencies": {
+    "@types/node": "^22.0.0",
+    "typescript": "^5.6.0"
+  }
+}

modules/systemd/tools/index.ts

@@ -0,0 +1,71 @@
+// systemd's tools: the node-service-manager seat's eight verbs — the units on this machine in
+// both scopes, read and acted on by name — and the module's own reading of what has failed
+// (novox/hq ADR 0177). Served by the node tools runtime (ADR 0175); the host applies units, this
+// answers about them.
+import { registerModuleTools, type ToolDefinition } from "@novox/mesh-sdk/tools";
+import { ServiceManager, type Scope } from "../client.js";
+
+const scope = { type: "string", description: "\"system\" (the default) or \"user\": the operator account's own manager" };
+const unit = { type: "string", description: "the unit's name, as the service manager knows it" };
+
+function scopeOf(args: Readonly<Record<string, unknown>>): Scope {
+  const s = String(args.scope ?? "system");
+  if (s !== "system" && s !== "user") throw new Error(`scope ${JSON.stringify(s)}: "system" or "user"`);
+  return s;
+}
+function unitOf(args: Readonly<Record<string, unknown>>): string {
+  const u = String(args.unit ?? "").trim();
+  if (!u) throw new Error("a unit is required");
+  return u;
+}
+
+export function getSeatVerbs(manager: ServiceManager): ToolDefinition[] {
+  const act = (verb: "start" | "stop" | "restart" | "enable" | "disable", description: string): ToolDefinition => ({
+    name: verb,
+    description,
+    input: { type: "object", properties: { scope, unit }, required: ["unit"] },
+    run: async (args) => manager.act(scopeOf(args), verb, unitOf(args)),
+  });
+  return [
+    {
+      name: "units",
+      description: "The units the service manager knows in a scope, each with its load, active and sub state; narrowed to a pattern when asked.",
+      input: { type: "object", properties: { scope, pattern: { type: "string", description: "a glob the unit's name must match (optional)" } } },
+      run: async (args) => ({ scope: scopeOf(args), units: await manager.units(scopeOf(args), args.pattern ? String(args.pattern) : undefined) }),
+    },
+    {
+      name: "status",
+      description: "One unit as the service manager sees it now: its states, whether it starts at boot, its main process, and whether the mesh declares it.",
+      input: { type: "object", properties: { scope, unit }, required: ["unit"] },
+      run: async (args) => manager.status(scopeOf(args), unitOf(args)),
+    },
+    act("start", "Start one unit. For a unit the mesh declares, the answer says the host will restore what its declaration says at the next apply."),
+    act("stop", "Stop one unit; for a mesh-declared unit the answer says the host will restore its declared state."),
+    act("restart", "Restart one unit."),
+    act("enable", "Make one unit start at boot (or at the account's login, in user scope)."),
+    act("disable", "Stop one unit starting at boot (or at login, in user scope)."),
+    {
+      name: "journal",
+      description: "The last lines of one unit's journal.",
+      input: { type: "object", properties: { scope, unit, lines: { type: "number", description: "how many lines from the end (default 100)" } }, required: ["unit"] },
+      run: async (args) => {
+        const n = Number(args.lines ?? 100);
+        return manager.journal(scopeOf(args), unitOf(args), Number.isFinite(n) && n > 0 ? Math.min(n, 5000) : 100);
+      },
+    },
+  ];
+}
+
+export function getOwnTools(manager: ServiceManager): ToolDefinition[] {
+  return [
+    {
+      name: "systemd_failed",
+      description: "Every failed unit on this machine, in the system manager and in the operator account's.",
+      input: { type: "object", properties: {} },
+      run: async () => manager.failed(),
+    },
+  ];
+}
+
+registerModuleTools("node-service-manager", (env) => getSeatVerbs(ServiceManager.fromEnv(env)));
+registerModuleTools("systemd", (env) => getOwnTools(ServiceManager.fromEnv(env)));

modules/systemd/tsconfig.json

@@ -0,0 +1,15 @@
+{
+  "compilerOptions": {
+    "target": "ES2022",
+    "module": "NodeNext",
+    "moduleResolution": "NodeNext",
+    "strict": true,
+    "esModuleInterop": true,
+    "skipLibCheck": true,
+    "noEmit": true
+  },
+  "include": [
+    "tools/index.ts",
+    "client.ts"
+  ]
+}

modules/zsh/module.json

@@ -0,0 +1,81 @@
+{
+  "module": "zsh",
+  "version": "1",
+  "capabilities": [
+    "package-manager"
+  ],
+  "seats": [
+    {
+      "name": "login-shell",
+      "scope": "node",
+      "serves": [
+        {
+          "name": "execute",
+          "description": "Run one command on this machine as the operator account, in a login shell; answers with what it printed and how it exited (novox/hq ADR 0176).",
+          "input": {
+            "type": "object",
+            "properties": {
+              "command": {
+                "type": "string",
+                "description": "the command line, as you would type it"
+              },
+              "timeout_seconds": {
+                "type": "number",
+                "description": "give up after this long (default 60)"
+              }
+            },
+            "required": [
+              "command"
+            ]
+          }
+        }
+      ]
+    }
+  ],
+  "claims": [
+    {
+      "name": "login-shell",
+      "scope": "node",
+      "serves": [
+        "execute"
+      ]
+    }
+  ],
+  "tools": [
+    "zsh_config"
+  ],
+  "resources": [
+    {
+      "id": "package",
+      "type": "package",
+      "package": "zsh"
+    },
+    {
+      "id": "rc",
+      "type": "file",
+      "path": "${machine:account-home}/.zshrc",
+      "owner": "${machine:account}",
+      "mode": "0644",
+      "into": "block",
+      "content": "# The mesh's default zsh configuration (module zsh). Everything OUTSIDE this block is yours and\n# survives every push; everything inside it is replaced on the next one (novox/hq ADR 0174).\n# Machine-specific lines go in ~/.zshrc.local, which this sources last.\n\nexport EDITOR=vim\nexport VISUAL=vim\nexport XDG_CONFIG_HOME=\"$HOME/.config\"\nexport PATH=\"$HOME/.local/bin:$HOME/scripts:$HOME/scripts/bin:$PATH\"\n\n# Terminal title: host, directory, git branch\nfunction set_terminal_title() {\n  local git_branch=\"\"\n  if git rev-parse --is-inside-work-tree &>/dev/null; then\n    git_branch=\" ($(git branch --show-current 2>/dev/null))\"\n  fi\n  print -Pn \"\\e]2;%m: %~${git_branch}\\a\"\n}\nprecmd_functions+=(set_terminal_title)\n\n# A prompt theme and plugins, when a module placed them (the prompt module owns ~/.p10k.zsh and\n# ~/.zsh/themes; this only loads what is there).\n[[ ! -f ~/.zsh/themes/powerlevel10k/powerlevel10k.zsh-theme ]] || source ~/.zsh/themes/powerlevel10k/powerlevel10k.zsh-theme\n[[ ! -f ~/.p10k.zsh ]] || source ~/.p10k.zsh\n[[ ! -f ~/.zsh/plugins/zsh-autosuggestions/zsh-autosuggestions.zsh ]] || source ~/.zsh/plugins/zsh-autosuggestions/zsh-autosuggestions.zsh\n[[ ! -f ~/.zsh/plugins/zsh-syntax-highlighting/zsh-syntax-highlighting.zsh ]] || source ~/.zsh/plugins/zsh-syntax-highlighting/zsh-syntax-highlighting.zsh\n\n# Keybindings: Home, End, Ctrl-A, Ctrl-E, Del\nbindkey \"^[[H\"  beginning-of-line\nbindkey \"^[OH\"  beginning-of-line\nbindkey \"^A\"    beginning-of-line\nbindkey \"^[[F\"  end-of-line\nbindkey \"^[OF\"  end-of-line\nbindkey \"^E\"    end-of-line\nbindkey \"^[[3~\" delete-char\n\n# Colour and the usual ls aliases\nif [ -x /usr/bin/dircolors ]; then\n  test -r \"$HOME/.dircolors\" && eval \"$(dircolors -b \"$HOME/.dircolors\")\" || eval \"$(dircolors -b)\"\n  alias ls='ls --color=auto'\n  alias grep='grep --color=auto'\nfi\nalias ll='ls -alhF'\nalias la='ls -Ah'\nalias l='ls -CFh'\nalias drun='docker run -it --rm'\ndisksize() { du -h --max-depth=1 \"${1:-.}\" | sort -h; }\n\n# Machine-specific configuration, kept by you\n[[ ! -f ~/.zshrc.local ]] || source ~/.zshrc.local\n"
+    },
+    {
+      "id": "login",
+      "type": "user",
+      "name": "${machine:account}",
+      "shell": "/usr/bin/zsh"
+    }
+  ],
+  "build": {
+    "artifacts": [
+      {
+        "name": "tools",
+        "kind": "bundle",
+        "language": "typescript",
+        "entrypoints": [
+          "tools/index.js"
+        ]
+      }
+    ]
+  }
+}

modules/zsh/package.json

@@ -0,0 +1,14 @@
+{
+  "name": "@novox/module-zsh",
+  "version": "0.1.0",
+  "description": "zsh \u2014 the shell as a module: the package, the mesh's default ~/.zshrc as a block the operator's own lines survive around, the login-shell seat and its execute verb (novox/hq ADR 0176).",
+  "type": "module",
+  "private": true,
+  "dependencies": {
+    "@novox/mesh-sdk": "^0.1.0"
+  },
+  "devDependencies": {
+    "@types/node": "^22.0.0",
+    "typescript": "^5.6.0"
+  }
+}

modules/zsh/tools/index.ts

@@ -0,0 +1,98 @@
+// zsh's tools — the module's own, and its implementation of the login-shell seat's one verb
+// (novox/hq ADR 0176). Served by the node tools runtime (ADR 0175); nothing here runs a process.
+//
+// `execute` runs as the operator account. The runtime runs as the node's account — root when the
+// host started it — so the command is handed to the account through `runuser` when we are not
+// already that account. Root is the module's concern (ADR 0175 §4): a command that needs it uses
+// sudo inside the shell like a person would.
+
+import { spawn } from "node:child_process";
+import { readFile } from "node:fs/promises";
+import { homedir, userInfo } from "node:os";
+import { registerModuleTools, type ToolDefinition } from "@novox/mesh-sdk/tools";
+
+/** The operator account on this machine, as the mesh told the runtime; the current user otherwise. */
+function account(env: NodeJS.ProcessEnv): string {
+  return env.MESH_OPERATOR_ACCOUNT?.trim() || userInfo().username;
+}
+
+interface Executed {
+  command: string;
+  account: string;
+  status: number | null;
+  signal: string | null;
+  stdout: string;
+  stderr: string;
+  timed_out: boolean;
+}
+
+/** Run one command line in a zsh login shell as the account, capturing everything. */
+export async function execute(command: string, who: string, timeoutSeconds: number): Promise<Executed> {
+  const self = userInfo().username;
+  const argv = who === self
+    ? ["zsh", "-lc", command]
+    : ["runuser", "-u", who, "--", "zsh", "-lc", command];
+  return new Promise((resolve) => {
+    const child = spawn(argv[0], argv.slice(1), { stdio: ["ignore", "pipe", "pipe"] });
+    let stdout = "";
+    let stderr = "";
+    let timedOut = false;
+    child.stdout.on("data", (d: Buffer) => { stdout += d.toString(); });
+    child.stderr.on("data", (d: Buffer) => { stderr += d.toString(); });
+    const timer = setTimeout(() => { timedOut = true; child.kill("SIGKILL"); }, timeoutSeconds * 1000);
+    child.on("error", (err) => {
+      clearTimeout(timer);
+      resolve({ command, account: who, status: null, signal: null, stdout, stderr: stderr + err.message, timed_out: false });
+    });
+    child.on("close", (status, signal) => {
+      clearTimeout(timer);
+      resolve({ command, account: who, status, signal, stdout, stderr, timed_out: timedOut });
+    });
+  });
+}
+
+function seatVerbs(env: NodeJS.ProcessEnv): ToolDefinition[] {
+  return [
+    {
+      name: "execute",
+      description: "Run one command on this machine as the operator account, in a login shell; answers with what it printed and how it exited.",
+      input: {
+        type: "object",
+        properties: {
+          command: { type: "string", description: "the command line, as you would type it" },
+          timeout_seconds: { type: "number", description: "give up after this long (default 60)" },
+        },
+        required: ["command"],
+      },
+      run: async (args) => {
+        const command = String(args.command ?? "").trim();
+        if (!command) throw new Error("execute: a command is required");
+        const timeout = Number(args.timeout_seconds ?? 60);
+        return execute(command, account(env), Number.isFinite(timeout) && timeout > 0 ? timeout : 60);
+      },
+    },
+  ];
+}
+
+function ownTools(env: NodeJS.ProcessEnv): ToolDefinition[] {
+  return [
+    {
+      name: "zsh_config",
+      description: "The operator account's ~/.zshrc on this machine as it is now: the mesh's block and the lines around it.",
+      input: { type: "object", properties: {} },
+      run: async () => {
+        const who = account(env);
+        const home = env.MESH_OPERATOR_HOME?.trim() || (who === userInfo().username ? homedir() : `/home/${who}`);
+        const path = `${home}/.zshrc`;
+        const text = await readFile(path, "utf8").catch(() => "");
+        const inBlock = /# BEGIN mesh [^\n]*\n([\s\S]*?)# END mesh/.exec(text);
+        return { account: who, path, lines: text.split("\n").length, mesh_block_lines: inBlock ? inBlock[1].split("\n").length - 1 : 0, content: text };
+      },
+    },
+  ];
+}
+
+// The seat's verb is registered under the seat's name (what the runtime serves on the seat's
+// subject when this module holds it) and the module's own tools under the module's.
+registerModuleTools("login-shell", seatVerbs);
+registerModuleTools("zsh", ownTools);

modules/zsh/tsconfig.json

@@ -0,0 +1,14 @@
+{
+  "compilerOptions": {
+    "target": "ES2022",
+    "module": "NodeNext",
+    "moduleResolution": "NodeNext",
+    "strict": true,
+    "esModuleInterop": true,
+    "skipLibCheck": true,
+    "noEmit": true
+  },
+  "include": [
+    "tools/index.ts"
+  ]
+}