菜单
Owns `infra/remotes/*.yaml` — fleet ops (inventory, WoL, SSH/RDP, reachability, shutdown/reboot) + service ops (health, logs, restart, deploy). "remote" = physical machine, not git remote. Triggers: "remote", "my PC", "my machines", "fleet", "wake", "WoL", "ssh to", "RDP", machine names (homeserver, workstation, laptop, router).
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Contribute features and improvements from your Bridge back to the open-bridge OSS upstream (or your org overlay) as a fork-based pull request. Scans your user branch for upstream-worthy files, classifies them by scope, adapts org-specific content, runs the MANDATORY two-layer content-safety gate (leak scanner + per-repo blocklist, refuses on PII/customer hits), and opens the cross-fork PR with DCO sign-off — no upstream push access needed. Trigger: "/contribute", "contribute", "upstream contribution", "contribute to upstream", "contribute upstream", "submit improvement", "share my skill", "send this to open-bridge", "PR to upstream", "PR to overlay", "give back", "submit a feature".
Generates the Bridge Control Center — a single-file HTML dashboard that bundles Fleet (infra/remotes/), Work Board, Calendar (next 24h), Channels, Git activity, optional Bridge-Deck /metrics, and upstream drift into one view. Stands alone; the pixel-art Bridge-Deck renderer it can complement is not yet public (coming soon). Trigger: "/bridge-dashboard", "bridge dashboard", "control center", "ops dashboard", "open dashboard", "show me everything at a glance".
Content-leak scanner with categorized classification. Different from a raw blocklist grep: each hit is sorted into legitimate-self-reference, legitimate-sister-repo, personal-PII (always leak), or internal-vocabulary hardcoded (OSS-strict only). Use after a sync to verify the destination is actually clean — catches pre-existing leaks the source-side scan won't see. Also runs OSS-strictness vocabulary checks (e.g. "scope: bks" hardcoded in OSS-shipped skill docs that should use "scope: org" placeholder). Trigger: "/bridge-leak-check", "leak scan", "leak check", "OSS readiness", "pre-existing leaks", "cleanliness check", "is this OSS clean".
New user onboarding and reconfiguration — discovery-driven setup with permission-gated system scan, evidence-based feature suggestions, and a read-only catalogue of what else Bridge can do. Six-phase wizard (Identity / Discovery / Suggestions / Quick-Wins / Catalog / Validate) with re-entry modes for targeted later activation. Works without GitHub. Upstreams stay empty by default and wire later when the OSS upstream or your own upstream is live. Trigger: "/bridge-onboard", "onboard", "setup", "configure bridge", "new user", "ecosystem scan", "set up bridge", "reconfigure", "reconfigure bridge", "setup wizard".
Subscribe a Bridge to an ORG OVERLAY — a git repo an organisation publishes that ships its shared config (contexts, projects, mandants, accounts, org skills/agents/standing-orders, an ecosystem fragment) as a flat mirror tree. /overlay add <git-url> sparse-clones the overlay, validates its manifest, previews a per-file plan with risk flags, and materializes COPIES into your repo under a lockfile — never block-merging your config, never clobbering your edits (3-way merge), never touching CORE. sync/apply/status/diff/remove/list manage the subscription over time. The overlay is the LOWER layer; your user files always win. Trigger: "/overlay", "org overlay", "subscribe to overlay", "materialize org config", "pull org overlay", "unsubscribe overlay", "add org config by git url", "list overlays", "overlay status", "overlay diff".
Promote CORE changes from your user branch to upstream (Scope-Routing per scope: core → OSS upstream, org → your org overlay, user → local). Analyzes commits, categorizes by CORE/USER/ORG path+frontmatter rules, runs mandatory content-safety checks per repo (leak scanner + blocklist, refuses on PII/customer hits), and creates fork-based PRs to upstream repos — no upstream push access needed. For file-level community contributions with adaptation, use the bridge-contribute skill (/contribute) instead. Trigger: "/bridge-promote", "promote", "cherry-pick", "push to upstream", "promote to upstream", "promote commits".
基于 SOC 职业分类
| name | remote |
| description | Owns `infra/remotes/*.yaml` — fleet ops (inventory, WoL, SSH/RDP, reachability, shutdown/reboot) + service ops (health, logs, restart, deploy). "remote" = physical machine, not git remote. Triggers: "remote", "my PC", "my machines", "fleet", "wake", "WoL", "ssh to", "RDP", machine names (homeserver, workstation, laptop, router). |
| metadata | {"scope":"core"} |
Manage remote machines defined in infra/remotes/*.yaml.
Read the referenced file ONLY when triggered.
remotes.enabled must be true in bridge-config.yaml. If not: inform and exit.
| Argument | Effect | Default |
|---|---|---|
(none) | Dashboard of all remotes | — |
{hostname} | Detail view of one remote | — |
health {hostname} | SSH health check | — |
logs {hostname} {service} | Show recent logs | — |
restart {hostname} {service} | Restart service | — |
sync {hostname} | Sync config files | — |
wake {hostname} | Wake-on-LAN | — |
connect {hostname} | Open SSH / RDP session | — |
shutdown {hostname} | Power off (requires [y]) | — |
reboot {hostname} | Restart machine (requires [y]) | — |
User wants to...
├── Remote overview / dashboard → Read references/workflow.md (§ Dashboard)
├── Detail view of one machine → Read references/workflow.md (§ Dashboard, single host)
├── Wake / power on a machine → Read references/fleet.md (§ Stage 3 WoL)
├── Connect via SSH / RDP → Read references/fleet.md (§ Stage 4 Connect)
├── Reboot or shutdown → Read references/fleet.md (§ Stage 5)
├── Check reachability / online? → Read references/fleet.md (§ Stage 2)
├── Update inventory after a change → Read references/fleet.md (§ Stage 6)
├── Health check for a machine → Read references/workflow.md (§ Health Check)
├── View service logs → Read references/workflow.md (§ Logs)
├── Restart a service → Read references/workflow.md (§ Restart)
├── Sync config to a remote → Read references/workflow.md (§ Sync)
├── Deploy services to a remote → Read references/workflow.md (§ Deploy)
├── Identify "my PC" / intent-only → Read references/fleet.md (§ Stage 1 + Learned routing)
└── Questions about remotes → Answer from CLAUDE.md § Remotes
| File | Owns |
|---|---|
references/fleet.md | Inventory, WoL, connect, shutdown, reachability, "my PC" routing |
references/workflow.md | Service dashboard, health, logs, restart, sync, deploy |
[y]infra/remotes/*.yaml — KeyVault / 1Password URIs onlywake_on_lan.enabled: false — never force wake a machine that opted outinfra/remotes/{target}.yaml before any operation — no guessing