Claude Skill

reboot

Reboot a fleet node safely by one protocol: pre-flight (save state, finish syncing, warn peers, verify autostart is armed), then a full restart rather than a fast-startup hybrid, then a post-reboot check through the crash-recovery skill. Triggers: "/reboot", "restart this machine

LLM Mart · 0 points · 1 views 0 listing impressions 0 install-command copies
Download tonydzi-second-brain-starter-kit-skills_reboot-7bc2efa.zip · 2 KB

Install

skills CLI npx skills add https://github.com/tonydzi/second-brain-starter-kit/tree/master/skills/reboot
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install tonydzi-second-brain-starter-kit@llmmart
Git git clone https://github.com/tonydzi/second-brain-starter-kit.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole tonydzi/second-brain-starter-kit collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

/reboot — a controlled restart of a fleet node

The pain: rebooting blind = a lost unsynced file (sync-conflict), a false SEV1 alarm from the peer watchdogs ("the node is dead!"), a robot that never came back after boot, and — as with the HP Wolf driver on 2026-07-24 — "I rebooted and the driver is still in memory", because Fast Startup does not clear RAM on an ordinary shutdown.

Reboot ONLY when: (a) there is a reason (evict a hook/driver from RAM, apply an update, unstick something), and (b) autostart is armed → the machine will come back on its own.

STEP 1 — PRE-FLIGHT (before the reboot, in order)

  1. Reason + confirmation. One line: why the reboot and what it fixes. If it is risky (away-mode, a machine without auto-login, FileVault/BitLocker with a manual password at boot) → ask the operator first.
  2. Save state. The TurnState black box writes on every turn by itself (it survives the reboot → /1 will bring it back). Any active uncommitted work in code or the vault → commit/save it now.
  3. Finish the Syncthing sync (otherwise: loss/conflict). Wait for need=0 on every share:
    powershell -NoProfile -ExecutionPolicy Bypass -File "$IMPORTS_ROOT/sync_check/sync_check.ps1"
    
    need>0 → wait for the sync / raise it (/raise-sync); do NOT reboot with a pending need.
  4. Warn the peers (otherwise the observers raise a false offline SEV1) — into the bus + the fleet log chat, BEFORE the reboot:
    python "$HOME/.claude/scripts/bus_send.py" --text "🔁 <host> going down for a reboot (~N min), reason: <...>. I'll come back on my own."
    
  5. Verify autostart is armed (the machine must come back WITHOUT hands):
    • PC: Claude Desktop = a single launcher %APPDATA%\...\Startup\Claude-Autostart.lnk (shell:AppsFolder\Claude_pzs8sxrjxfjjc!Claude), NOT a broken HKCU\Run pointing at an old version ([[claude-desktop-autostart-race]]); Syncthing = start-syncthing.vbs in Startup; the watchdog tasks in place. On the hub, also auto-login + hub_boot_report.py ([[hub-boot-selfreport]]).
    • Mac: Login Items / launchd agents for Syncthing + Claude; a FileVault password at boot is manual (without it an unattended reboot will not come back — only do it with a human present).

STEP 2 — THE REBOOT (with the right command)

⚠️ A FULL Restart, not a Shutdown — Fast Startup (hiberboot) on Windows does NOT clear the kernel/drivers/RAM on shutdown /s; Restart (/r) always does a full cycle and clears them. To "evict a driver/hook from memory" (HP Wolf, an antivirus, a stuck driver) ONLY a full Restart works.

  • PC (Windows):
    shutdown /r /t 0
    
    (a guaranteed clean exit even with Fast Startup ON; the Claude Code session ends with it).
  • Mac:
    osascript -e 'tell app "System Events" to restart'   # or: sudo shutdown -r now
    

STEP 3 — POST-REBOOT (once it's up — through /1)

The machine is back → in a new session run /1 (resurrection): RECALL from the black box + a health ping (arch/sync/mcp) + the previous session's full history into the buffer. Then:

  • Confirm autostart fired: Syncthing connected, Claude Desktop = a single instance (no race), connectors green.
  • Check that the reason for the reboot was actually achieved (e.g. the black windows are gone / the driver is no longer in memory) — by looking or measuring, not on faith ([[prichina-kak-claim]]).
  • A boot self-report into the bus (the hub does it itself; on a peer — a short "✅

Boundaries

  • READ-only right up to the reboot itself; the reboot is a deliberate action with a stated reason.
  • ⛔ Away-mode / no auto-login / a manual FileVault-BitLocker prompt at boot → reboot only with the operator's approval (risk: it never comes back without hands).
  • A peer reboots ITSELF ([[peers-own-outbound-local-full-member]]); rebooting the hub while away — carefully ([[away-mode-45-days]], BIOS "Restore on AC Power Loss").
  • The folder name is reboot; /restart is a text trigger for the same skill (case does not matter).

Like this skill? It is part of second-brain-starter-kit: the second brain we built for ourselves and run every day at Palo Alto AI Research Lab. Install the whole set with npx skills add tonydzi/second-brain-starter-kit. Everything is open source and free, so take what you need.

Flagships worth a look on their own: secondop-panel (a second opinion from a panel of external models), claude-memory-tidy (stop your agent's memory from rotting), telegram-mcp-kit (your own Telegram over MCP in about 15 minutes).

Author: Anton Dziatkovskii, Palo Alto AI Research Lab. Telegram @tonydzi - WhatsApp +1 341 222 9178 - X @Tony_Stef_

Engineers: want to test-drive this setup? Message me. I hand out free starter seeds to engineers who test and report back, and custom skill requests are welcome.

Files (second-brain-starter-kit)
  • SKILL.md 5.5 KB
    ---
    name: reboot
    description: >-
      Reboot a fleet node safely by one protocol: pre-flight (save state, finish syncing, warn
      peers, verify autostart is armed), then a full restart rather than a fast-startup hybrid, then
      a post-reboot check through the crash-recovery skill. Triggers: "/reboot", "restart this
      machine".
    license: MIT
    ---
    
    # /reboot — a controlled restart of a fleet node
    
    **The pain:** rebooting blind = a lost unsynced file (sync-conflict), a false SEV1 alarm from the
    peer watchdogs ("the node is dead!"), a robot that never came back after boot, and — as with the
    HP Wolf driver on 2026-07-24 — "I rebooted and the driver is still in memory", because Fast Startup
    does not clear RAM on an ordinary shutdown.
    
    Reboot ONLY when: (a) there is a reason (evict a hook/driver from RAM, apply an update, unstick
    something), and (b) autostart is armed → the machine will come back on its own.
    
    ## STEP 1 — PRE-FLIGHT (before the reboot, in order)
    
    1. **Reason + confirmation.** One line: why the reboot and what it fixes. If it is risky
       (away-mode, a machine without auto-login, FileVault/BitLocker with a manual password at boot) → ask the operator first.
    2. **Save state.** The TurnState black box writes on every turn by itself (it survives the reboot → /1 will bring it back).
       Any active uncommitted work in code or the vault → commit/save it now.
    3. **Finish the Syncthing sync** (otherwise: loss/conflict). Wait for `need=0` on every share:
       ```bash
       powershell -NoProfile -ExecutionPolicy Bypass -File "$IMPORTS_ROOT/sync_check/sync_check.ps1"
       ```
       need>0 → wait for the sync / raise it (`/raise-sync`); do NOT reboot with a pending need.
    4. **Warn the peers** (otherwise the observers raise a false offline SEV1) — into the bus + the fleet log chat, BEFORE the reboot:
       ```bash
       python "$HOME/.claude/scripts/bus_send.py" --text "🔁 <host> going down for a reboot (~N min), reason: <...>. I'll come back on my own."
       ```
    5. **Verify autostart is armed** (the machine must come back WITHOUT hands):
       - **PC:** Claude Desktop = a single launcher `%APPDATA%\...\Startup\Claude-Autostart.lnk`
         (`shell:AppsFolder\Claude_pzs8sxrjxfjjc!Claude`), NOT a broken HKCU\Run pointing at an old version
         ([[claude-desktop-autostart-race]]); Syncthing = `start-syncthing.vbs` in Startup; the watchdog tasks in place.
         On the hub, also auto-login + `hub_boot_report.py` ([[hub-boot-selfreport]]).
       - **Mac:** Login Items / launchd agents for Syncthing + Claude; a FileVault password at boot is manual
         (without it an unattended reboot will not come back — only do it with a human present).
    
    ## STEP 2 — THE REBOOT (with the right command)
    
    ⚠️ **A FULL Restart, not a Shutdown** — Fast Startup (hiberboot) on Windows does NOT clear the
    kernel/drivers/RAM on `shutdown /s`; `Restart` (`/r`) always does a full cycle and clears them. To
    "evict a driver/hook from memory" (HP Wolf, an antivirus, a stuck driver) ONLY a full Restart works.
    
    - **PC (Windows):**
      ```bash
      shutdown /r /t 0
      ```
      (a guaranteed clean exit even with Fast Startup ON; the Claude Code session ends with it).
    - **Mac:**
      ```bash
      osascript -e 'tell app "System Events" to restart'   # or: sudo shutdown -r now
      ```
    
    ## STEP 3 — POST-REBOOT (once it's up — through /1)
    
    The machine is back → in a new session run **`/1`** (resurrection): RECALL from the black box + a
    health ping (arch/sync/mcp) + the previous session's full history into the buffer. Then:
    - Confirm autostart fired: Syncthing connected, Claude Desktop = a single instance (no race),
      connectors green.
    - **Check that the reason for the reboot was actually achieved** (e.g. the black windows are gone /
      the driver is no longer in memory) — by looking or measuring, not on faith ([[prichina-kak-claim]]).
    - A boot self-report into the bus (the hub does it itself; on a peer — a short "✅ <host> is back, all green").
    
    ## Boundaries
    - READ-only right up to the reboot itself; the reboot is a deliberate action with a stated reason.
    - ⛔ Away-mode / no auto-login / a manual FileVault-BitLocker prompt at boot → reboot only with the operator's approval
      (risk: it never comes back without hands).
    - A peer reboots ITSELF ([[peers-own-outbound-local-full-member]]); rebooting the hub while away — carefully
      ([[away-mode-45-days]], BIOS "Restore on AC Power Loss").
    - The folder name is `reboot`; `/restart` is a text trigger for the same skill (case does not matter).
    
    <!--kit-footer-->
    
    ---
    
    **Like this skill?** It is part of [second-brain-starter-kit](https://github.com/tonydzi/second-brain-starter-kit): the second brain we built for ourselves and run every day at Palo Alto AI Research Lab. Install the whole set with `npx skills add tonydzi/second-brain-starter-kit`. Everything is open source and free, so take what you need.
    
    Flagships worth a look on their own: [secondop-panel](https://github.com/tonydzi/secondop-panel) (a second opinion from a panel of external models), [claude-memory-tidy](https://github.com/tonydzi/claude-memory-tidy) (stop your agent's memory from rotting), [telegram-mcp-kit](https://github.com/tonydzi/telegram-mcp-kit) (your own Telegram over MCP in about 15 minutes).
    
    Author: **Anton Dziatkovskii**, Palo Alto AI Research Lab. Telegram [@tonydzi](https://t.me/tonydzi) - WhatsApp [+1 341 222 9178](https://wa.me/13412229178) - X [@Tony_Stef_](https://x.com/Tony_Stef_)
    
    **Engineers: want to test-drive this setup?** Message me. I hand out free starter seeds to engineers who test and report back, and custom skill requests are welcome.
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.