AGENT SKILL

ccb-mobile-relay-maintain

Diagnose and safely maintain the CCB Mobile host service, LAN/tailnet/Cloudflare Tunnel/Relay routes, pairing and devices, Relay invitation or host activation, multi-project terminals, and capability-negotiated Provider controls. Use when Mobile cannot connect, pair, discover projects, open terminals, expose Provider controls, or when a user explicitly asks to activate, update, revoke, or repair a CCB Relay host or device.

License 许可
Apache-2.0
最近更新
2026-08-14
Stars
★ 19
Forks
2

01 前言 Front Matter

name
ccb-mobile-relay-maintain
description
Diagnose and safely maintain the CCB Mobile host service, LAN/tailnet/Cloudflare Tunnel/Relay routes, pairing and devices, Relay invitation or host activation, multi-project terminals, and capability-negotiated Provider controls. Use when Mobile cannot connect, pair, discover projects, open terminals, expose Provider controls, or when a user explicitly asks to activate, update, revoke, or repair a CCB Relay host or device.

03 SKILL.md 正文

CCB Mobile Relay Maintain

Read references/ccb-mobile-relay-runtime.md before mutation. Keep the Mobile host's server-wide authority separate from project runtime and Pane evidence.

Safety Rules

  • Never read, print, store, or request Relay credentials, invitation secrets, device secrets, Provider credentials, or raw Provider responses in chat or Role memory.
  • Diagnose before replacing the host or rotating pairing state.
  • Revoke only one exact device, invitation, or host after explicit user intent.
  • Do not infer Mobile host health from a project Pane or .ccb/agents/*.
  • Do not force Provider controls when capability negotiation says unsupported; report the required host update.

Workflow

  1. Classify the incident:
    • host service not running or unhealthy;
    • LAN/tailnet/tunnel/Relay route unreachable;
    • pairing handoff expired or unclaimable;
    • device unauthorized or intentionally revoked;
    • running project not discovered;
    • host or project terminal transport failed;
    • Provider controls hidden, rejected, or pending restart;
    • native Windows registry, endpoint, TCP descriptor, or readiness mismatch.
  2. Collect sanitized evidence with the smallest applicable commands:
ccb mobile devices
ccb relay host status <host_id>
ccb relay host list
ccb relay invite status <invite_id>

Use local health and managed-service diagnostics when available. Do not expose secret-bearing files or arguments. 3. Verify the active service generation, CCB ownership of the listen endpoint, selected route provider, registered projects, and advertised capabilities. 4. Choose the least disruptive supported action:

  • ccb update mobile to refresh the server-wide host or rotate an expired pairing handoff;
  • ccb mobile revoke <device_id> for one explicitly selected device;
  • ccb relay host activate ... only from an operator-provided invitation or invitation file already available through the authorized workflow;
  • exact ccb relay ... revoke only for an explicitly selected identifier.
  1. For Provider setting changes, distinguish configured, active, and pending state. Hand guarded post-config Agent restart to ccb-self-recover when the host says restart is required.
  2. Verify health, route, project discovery, terminal connection, and capability negotiation after mutation.

Native Windows

Use native Windows process ownership, registry, TCP descriptor, endpoint marker, readiness, and Herdr namespace evidence. Do not apply Unix tmux or /proc assumptions. If the current release reports beta prerequisites or an unsigned launcher boundary, preserve that status in the report.

Output

Report:

  • incident class and current route;
  • sanitized host generation/health evidence;
  • project discovery and terminal status;
  • capability status and any host-update guidance;
  • exact mutation performed, if any;
  • identifiers revoked without secret material;
  • remaining platform or external-network blocker.

04 同仓库更多 Skills