[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77KQWZ7VTASPAH8D6NVES35":3,"skill-same:01M77KQWZ7VTASPAH8D6NVES35":57},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":29,"commitSha":30,"skillDirectory":31,"forksCount":32,"repositoryHtmlUrl":33,"repositoryHomepageUrl":6,"ownerHtmlUrl":34,"ownerBlogUrl":35,"ownerDescription":36,"updatedAt":37,"install":38,"resources":50},"01M77KQWZ7VTASPAH8D6NVES35","reach",null,"Reach another Claude Code lane in the fleet, on any machine, to run, prompt, query, message or start a session. Use when: 'run this on HOST', 'ask the desktop to', 'cross-machine', 'remote agent', 'reach the fleet', 'message the Windows session', 'from WSL to Windows', 'share one account across lanes'. Not for: a session in THIS lane (ListAgents\u002FSendMessage), a detached local background session (session-flow:continue-in-background), or repository fleets (repo-fleet-hygiene).","melodic-software\u002Fclaude-code-plugins","melodic-software","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F58273638?v=4","01M77JVQTXNGKCE4MDFY2BC936","Melodic Software","01M77JVQVN9QDWKAH6W03ANP76",30,"MIT","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Ffleet\u002Fskills\u002Freach\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\ndescription: \"Reach another Claude Code lane in the fleet, on any machine, to run, prompt, query, message or start a session. Use when: 'run this on HOST', 'ask the desktop to', 'cross-machine', 'remote agent', 'reach the fleet', 'message the Windows session', 'from WSL to Windows', 'share one account across lanes'. Not for: a session in THIS lane (ListAgents\u002FSendMessage), a detached local background session (session-flow:continue-in-background), or repository fleets (repo-fleet-hygiene).\"\nargument-hint: \"[relay]\"\nuser-invocable: true\ndisable-model-invocation: false\nmetadata:\n  workflow-stage: anytime\n  summary: Reach another fleet lane (WSL or Windows, here or remote) to run, prompt, query or message\n---\n\n## Resolve the target first\n\nA **lane** is one Claude Code install with its own sign-in: the WSL distro or native Windows, on\neach machine.\n\n1. Read `~\u002F.config\u002Ffleet\u002FFLEET.md` before composing anything. It holds the real aliases, accounts\n   and paths for this machine; the commands below use placeholders. Done when you can name the\n   target host and lane and see its reach entries (alias, port, account, shell).\n2. Pick the lane by the work. **WSL is the default agent lane** (repos, toolchain, sshd); use\n   Windows only for Windows-only work or a Windows-side session. Done when you hold a matrix row\n   with every placeholder filled.\n3. A host absent from FLEET.md is not a fleet host: say so and stop. Address machines by hostname\n   over the tailnet, never by session name.\n\n## Route matrix\n\n`\u003Cagent>` is the headless turn from [Verbs](#verbs): `claude` in a WSL lane, `\u003Cclaude-exe>` in a\nWindows lane. \"Tested\" means a real run of that exact shape; what each run showed is in\n[reference\u002Frelay.md](reference\u002Frelay.md#verified-behavior). The R2 and R3 runs passed the prompt\nas an argument; their stdin form below has not been re-probed.\n\n| # | From | To | Command shape | Status |\n|---|---|---|---|---|\n| R0 | any | same lane | Built-in `ListAgents` \u002F `SendMessage`. Not this skill | n\u002Fa |\n| R1 | WSL | this machine, Windows | `cd \u003Cwin-dir> && \u003Cclaude-exe> \u003Cagent-args> \"\u003Cprompt>\" \u003C \u002Fdev\u002Fnull` | tested: script (`powershell.exe`), list, multi-turn, message a `-n` receiver |\n| R2 | WSL | other machine, WSL | `printf '%s\\n' '\u003Cprompt>' \\| \u003Cssh> \u003Cwsl-alias> 'claude \u003Cagent-args>'` | tested: script, multi-turn, stream-json, `--bg`, message an interactive session and a `-p` receiver across accounts |\n| R3 | WSL | other machine, Windows | `printf '%s\\n' '\u003Cprompt>' \\| \u003Cssh> \u003Cwsl-alias> 'cd \u003Cwin-dir> && \u003Cclaude-exe> \u003Cagent-args>'` | reaches `claude.exe`; needs `\u002Flogin` at that machine's console |\n| R4 | Windows | this machine, WSL | `wsl.exe -d \u003Cdistro> --cd \u002Ftmp --exec \u003Cshell> -lc 'claude \u003Cagent-args> \"\u003Cprompt>\" \u003C \u002Fdev\u002Fnull'` | tested: script from a native-Windows `claude.exe` (its Bash tool, Git Bash); list |\n| R5 | Windows | other machine, WSL | `printf '%s\\n' '\u003Cprompt>' \\| ssh.exe \u003Cwsl-alias> 'claude \u003Cagent-args>'` | untested from a Windows origin (same hop as R2) |\n| R6 | Windows | other machine, Windows | `printf '%s\\n' '\u003Cprompt>' \\| ssh.exe \u003Cwsl-alias> 'cd \u003Cwin-dir> && \u003Cclaude-exe> \u003Cagent-args>'` | untested from a Windows origin (same hop as R3) |\n\nOver SSH a `-p` turn's prompt goes on stdin, and the quoted remote command carries only the flags,\nso no shell on either host parses prompt text. Never add `ssh -n` to an SSH row: it replaces stdin\nwith `\u002Fdev\u002Fnull` and drops the prompt. The local hops (R1, R4) keep the prompt as the last\nargument. A verb with no `\u003Cprompt>` drops the `printf` on an SSH row and the quoted prompt on R1\nand R4. `--bg` is not `-p` and takes its prompt as an argument, so a background session's prompt\nstays inside the remote command. In the tables here and in FLEET.md, `\\|` is an escaped `|`: type a bare `|`.\n\nA script with no agent goes over the same hop with the command in place of the `claude` part; on\nR1 that is `powershell.exe -NoProfile -Command '\u003Ccmd>'` from a Windows-side directory. A Windows\nscript on another machine goes over port 22 instead: `\u003Cssh> \u003Cwin-alias> '\u003Cpwsh command>'` (see\n[Port 22](#port-22-is-not-an-agent-lane)).\n\nPlaceholders, each filled from FLEET.md or a probe, never guessed:\n\n- `\u003Cssh>`: the Windows OpenSSH client. `ssh.exe` from Windows; from WSL,\n  `\u002Fmnt\u002Fc\u002FWindows\u002FSystem32\u002FOpenSSH\u002Fssh.exe`, because the distro holds no key.\n- `\u003Cwsl-alias>` \u002F `\u003Cwin-alias>`: the target's `wsl-shell` (port 2222) and `windows-shell` (port 22)\n  aliases.\n- `\u003Cclaude-exe>`: `\u002Fmnt\u002Fc\u002FUsers\u002F\u003Cuser>\u002F.local\u002Fbin\u002Fclaude.exe`, by absolute path.\n- `\u003Cwin-dir>`: a scratch directory under `\u002Fmnt\u002Fc\u002FUsers\u002F\u003Cuser>\u002F` (`claude.exe` needs a Windows-side\n  working directory), never a repository that defines the fleet's accounts or firewall.\n- Use FLEET.md's `\u002Fmnt\u002Fc` paths as given: never check them with file tools (Read, Glob, Grep),\n  and never ask a subagent's brief to. The hop check in\n  [Verify the hop](#verify-the-hop-before-you-trust-it) is what proves them.\n- `\u003Cdistro>`: from `wsl.exe -l -q`. `\u003Cshell>`: the distro account's login shell, so `claude` is on\n  `PATH`.\n\nPer-route detail, the receiver's nested quoting, reply patterns, background sessions, and the\nmechanisms not used here (Remote Control, cloud sessions, Channels) are in\n[reference\u002Frelay.md](reference\u002Frelay.md).\n\nConverging dotfiles on any fleet lane is not a route in this matrix: run `fleet-converge \u003Clane>`,\nby its bare name. It does not prompt because dotfiles renders an exact allow rule for that command\non every lane; no `Bash(ssh …)` rule is added, and the posture below stands for every other verb.\n\n- **Pointer**: for what `fleet-converge` checks and how it reaches each lane, read dotfiles\n  [docs\u002Fdesign-decisions.md, \"Converging a lane: `fleet-converge`\"](https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fdotfiles\u002Fblob\u002Fmain\u002Fdocs\u002Fdesign-decisions.md#converging-a-lane-fleet-converge).\n\n## Verify the hop before you trust it\n\nRun the row with `hostname &&` before the agent, `-p` as `\u003Cagent-args>` and `echo ok` as\n`\u003Cprompt>`, for example `printf '%s\\n' 'echo ok' | \u003Cssh> \u003Cwsl-alias> 'hostname && claude -p'`: the\nhostname proves which machine answered, `ok` that its Claude is signed in. A Windows lane answering with an\nexpired-OAuth-session authentication error routed correctly and is signed out: someone runs\n`\u002Flogin` at that machine's console (or over RDP). Nothing remote fixes it, and no credential\ntravels in a prompt.\n\n- **Pointer**: when matching a lane's answer to a signed-out state, fetch\n  [Login expired](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Ferrors#login-expired) live.\n\n## Verbs\n\nEach verb fills `\u003Cagent-args>` and `\u003Cprompt>` in a matrix row. Every verb but the first starts an\nagent, and each agent turn costs real usage, even a one-line turn\n([Manage costs: Track your costs](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcosts#track-your-costs); the Cost\nrow of [reference\u002Frelay.md § Verified behavior](reference\u002Frelay.md#verified-behavior) holds the\ndated measurement and how to re-measure it). When a script can do the job, run the script.\n\n| Verb | `\u003Cagent-args>` | `\u003Cprompt>` | Status |\n|---|---|---|---|\n| Run a script | none: the command itself replaces `claude ...` over the same hop | none | tested on R1, R2, R4; port 22 untested |\n| Prompt | `-p` | the task | tested on R1, R2, R4 |\n| Multi-turn | `-p --session-id \u003Cuuid>`, then `-p --resume \u003Cuuid>` against the same lane | each turn's prompt | tested on R1, R2 |\n| One open pipe | `-p --input-format stream-json --output-format stream-json --verbose` | none: user messages as NDJSON on stdin | tested on R2 |\n| List sessions | `-p` | `List the sessions you can reach` | tested on R1, R4 |\n| Message a session | `-p` | `SendMessage to \u003Cname>: \u003Ctext>` | tested on R1, R2 |\n| Named receiver | `-p -n \u003Cname> --settings '{\"crossSessionInbound\":\"accept\"}'` | the standing instructions | tested on R1, R2 |\n| Background session | `--bg --name \u003Cname> \"\u003Cprompt>\"`, not `-p`, with `\u003C \u002Fdev\u002Fnull` on an SSH row; manage with `claude agents --json --all`, `stop \u003Cid>`, `rm \u003Cid>` | none: it is in `\u003Cagent-args>` | tested on R2, trusted directory only |\n\n- Give every prompt the whole task: what done looks like, and what should make it stop and report.\n  Nobody answers a question a headless turn asks.\n- **Query and wait has no one-turn form.** Never wait on a `notify_when_idle` notice from a `-p`\n  sender: in our probe it never reached the sender\n  ([reference\u002Frelay.md § Verified behavior](reference\u002Frelay.md#verified-behavior)). Send the\n  message, then poll `claude agents --json` in that lane, or `--resume` the receiver or read its\n  output.\n  - **Pointer**: for how an idle notice is delivered and what a `-p` session can receive, fetch\n    [Get a notice when another session goes idle](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#get-a-notice-when-another-session-goes-idle)\n    and [Non-interactive sessions](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#non-interactive-sessions)\n    live.\n- **Multi-turn versus one open pipe.** The pipe keeps context in one process while the connection\n  stays open; per-turn `--resume` survives a disconnect. `--resume` ids belong to the lane that\n  made them, so resume against the same row. `--session-id` picks the id up front, so nothing\n  parses the first turn's output. Generate no id in this session: a command you hand over keeps\n  `\u003Cuuid>` for whoever runs it to fill, and a turn you run yourself can start with\n  `--output-format json` and resume with the `session_id` it returns (see \"Parse the JSON\" in\n  [Gotchas](#gotchas)). Say with the handed-over commands that the id exists only on that lane's\n  machine and account, so every follow-up goes to the same row.\n- **Background sessions** need a trusted working directory; anywhere else the start fails with a\n  workspace-not-trusted error and exit 1 (for its wording and fix, fetch\n  [Workspace not trusted when dispatching a background session](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Ferrors#workspace-not-trusted-when-dispatching-a-background-session)\n  live; as of 2026-10-11; recheck when that entry's message or fix changes). Address\n  `claude logs` and `claude attach` by the `id` from `claude agents --json --all` (bare `--json`\n  lists only active sessions); pass a name only after checking the lane's version against the CLI\n  reference. `logs` prints raw TUI output, so read a reply from the transcript (`--resume \u003Cid>`).\n  - **Pointer**: when deciding whether a lane can take a session name in place of the id, fetch\n    [CLI reference: CLI commands](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcli-reference#cli-commands) live.\n- A `-p` receiver needs `crossSessionInbound: accept`. For a session that must stay up, use a\n  background session, not a `-p` receiver. Give either one standing instructions as its prompt:\n  what to do with each message and when to stop. Delivery rules and the pointer are in\n  [reference\u002Frelay.md](reference\u002Frelay.md#a-named-receiver).\n  - **Pointer**: for when a `-p` run ends after its turn, fetch\n    [Run Claude Code programmatically: Background tasks at exit](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fheadless#background-tasks-at-exit)\n    live.\n- Use `--bare` for prompts, never for a receiver: a bare session cannot receive messages or be\n  listed.\n- Remote Control is interactive only: `-p --remote-control` does not connect. Keep it out of\n  headless recipes.\n- Session names belong to the target lane, and a listing shows only its own lane. List first, then\n  message a name from the listing that shows it. When a request names a machine and not a lane,\n  list each of that machine's lanes (on another machine, one R2 and one R3 turn).\n- **Pointer**: the receiver, `--bare` and Remote Control bullets rest on our probes and the flag\n  behavior in [reference\u002Frelay.md § Verified behavior](reference\u002Frelay.md#verified-behavior); when\n  one stops holding, fetch\n  [Control inbound messages](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#control-inbound-messages),\n  [Non-interactive sessions](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#non-interactive-sessions),\n  [Start faster with bare mode](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fheadless#start-faster-with-bare-mode)\n  and [CLI flags](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcli-reference#cli-flags) live.\n\n## Accounts are per lane\n\nEach lane signs into its own claude.ai account so one lane's usage never draws down another's (on\nthe desktop, WSL and Windows differ; the laptop's WSL lane uses a third). So:\n\n- **Same lane.** Sessions find each other through files and sockets; the peer tools work.\n- **Other lane, same machine.** WSL and Windows register under different homes and socket types;\n  each lane's `ListAgents` shows only its own lane.\n- **Other machine.** Remote Control lists only the signed-in account's sessions; across accounts\n  `ListAgents` shows nothing remote, even with Remote Control connected.\n\nThe peer tools are same-lane only, and no setting widens them; every other route starts a turn\ninside the target lane. Do not propose sharing an account: the split is the decision.\n\n- **Pointer**: the three facts are our probes' results\n  ([reference\u002Frelay.md § Verified behavior](reference\u002Frelay.md#verified-behavior)); when one is in\n  doubt, fetch\n  [See which sessions Claude can reach](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#see-which-sessions-claude-can-reach)\n  and\n  [Message sessions on other machines](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#message-sessions-on-other-machines)\n  live.\n\n## Port 22 is not an agent lane\n\n`ssh-admin` is a separate Windows account with no Claude sign-in and no DPAPI. Use port 22 for pwsh\nscripts only, with pwsh quoting; anything needing the console user's credentials goes through the\non-demand tasks FLEET.md lists.\n\n## Permission and safety posture\n\n- Every verb except running a script starts an agent in another lane, same machine included, and\n  these prompt under auto mode by design; let them prompt, and say so with every such command you\n  run, delegate or hand over. A script is judged on its own content.\n  When a prompt's cause is in doubt, fetch\n  [Eliminate prompts with auto mode](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fpermission-modes#eliminate-prompts-with-auto-mode)\n  live.\n- There is no `Bash(ssh ...)` allow rule anywhere in this fleet, and adding one is not the fix for a\n  prompt. If asked to stop the prompting, say what the rule would cost and decline.\n- Neither side runs `bypassPermissions`.\n- Never `wsl --shutdown` on a target, and never run `wsl -d` on a target's drift-convergence path:\n  both take the distro out from under whatever else uses it. The one exception is a start\n  `fleet-converge` makes itself, under the guard its pointer names; never run `wsl -d` on its\n  behalf.\n- On R1 and R4 the prompt travels in the command line, visible to the process table and shell\n  history; over SSH it travels on stdin, and the local shell history still records it. No secret\n  belongs in one.\n\n## Boundary\n\n| Neighbor | Owns |\n|---|---|\n| This skill | Every lane other than this session's: the other lane on this machine, and both lanes on other machines |\n| Built-in `ListAgents` \u002F `SendMessage` | Sessions in this lane, under this lane's account |\n| Built-in Remote Control | Human use of a lane's session from a phone or the web, under that lane's account. Not agent-to-agent across accounts |\n| `session-flow:continue-in-background` | A detached session in THIS lane |\n| `session-flow:orchestrate` | Delegation inside one session, to subagents |\n| `repo-fleet-hygiene:audit` | Fleets of REPOSITORIES. Same word, different subject |\n\n## Gotchas\n\n- **Use the Windows OpenSSH client.** It is agent-backed and is what `~\u002F.ssh\u002Fconfig` is wired to.\n  Git Bash's MSYS `ssh` reaches no agent and fails with `Permission denied (publickey)`, which reads\n  like a key problem and is not one.\n- **Give every `claude -p` turn a closed stdin.** Over SSH the piped prompt is stdin, and it ends\n  when `printf` does. On R1 and R4 redirect `\u003C \u002Fdev\u002Fnull`. A `-p` run reads stdin as its input, so\n  with no prompt argument the piped text is the prompt: see \"Pipe data through Claude\" and the\n  stdin-only example under \"Customize the system prompt\" in\n  [Run Claude Code programmatically](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fheadless#pipe-data-through-claude),\n  and the piped-content row of [CLI commands](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcli-reference#cli-commands).\n- **Escape apostrophes in `\u003Cprompt>`.** Every SSH row single-quotes the prompt in\n  `printf '%s\\n' '\u003Cprompt>'`, and R4 single-quotes the whole command, on the LOCAL shell, so a `'`\n  in the prompt (\"what's\") closes that quote early. The rule follows the origin shell, not the\n  lane:\n  - **bash, zsh, and Git Bash** (a native-Windows Claude's Bash tool, R4 to R6): replace each `'`\n    with `'\\''`, or wrap the prompt argument (on R4, the command) in `$'...'` and write `\\'`. Doubling (`''`) silently drops the\n    apostrophe here: `'what''s'` arrives as `whats`.\n  - **PowerShell**, Windows PowerShell 5.1 or pwsh 7 (a native-Windows Claude's PowerShell tool,\n    or a PowerShell window): on R4, keep the prompt in doubled single quotes (`-p ''...''`) and\n    write an apostrophe as `''\\''''`. Never use `-p \"...\"` from PowerShell, with or without a\n    doubled `''` apostrophe. This form ran correctly on every version probed. Run the SSH rows\n    from bash, zsh or Git Bash, whose `printf` is a builtin.\n    - **Probe**: [PowerShell quoting to wsl.exe](https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fissues\u002F12908#issuecomment-6624493777).\n      **As of**: 2026-11-04. **Recheck trigger**: the\n      [about_Parsing](https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fpowershell\u002Fmodule\u002Fmicrosoft.powershell.core\u002Fabout\u002Fabout_parsing?view=powershell-7.6#passing-arguments-that-contain-quote-characters)\n      page changes how quotes pass to native commands.\n- **Use `wsl.exe --exec`, not `--`.** The quoting rules above assume `--exec`. For what each option\n  does, run `wsl.exe --help` on that machine, as\n  [Basic commands: Help command](https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fwindows\u002Fwsl\u002Fbasic-commands#help-command)\n  directs (the page has no section of its own for `--exec`).\n- **Probe interop by absolute path.** `cmd.exe \u002Fc` prints nothing inside an sshd session and reads\n  as \"interop is broken\" when it is not; run the `.exe` by absolute path.\n- **Parse the JSON, not the stream.** Treat stdout as able to carry lines that are not the JSON\n  result, and extract the object first, for example `grep '^{' | jq -r .session_id`. What a\n  session-end hook can print is in [Hooks: SessionEnd](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fhooks#sessionend).\n- **A session name is not an address.** Hostname and lane, always.\n","## Resolve the target first\n\nA **lane** is one Claude Code install with its own sign-in: the WSL distro or native Windows, on\neach machine.\n\n1. Read `~\u002F.config\u002Ffleet\u002FFLEET.md` before composing anything. It holds the real aliases, accounts\n   and paths for this machine; the commands below use placeholders. Done when you can name the\n   target host and lane and see its reach entries (alias, port, account, shell).\n2. Pick the lane by the work. **WSL is the default agent lane** (repos, toolchain, sshd); use\n   Windows only for Windows-only work or a Windows-side session. Done when you hold a matrix row\n   with every placeholder filled.\n3. A host absent from FLEET.md is not a fleet host: say so and stop. Address machines by hostname\n   over the tailnet, never by session name.\n\n## Route matrix\n\n`\u003Cagent>` is the headless turn from [Verbs](#verbs): `claude` in a WSL lane, `\u003Cclaude-exe>` in a\nWindows lane. \"Tested\" means a real run of that exact shape; what each run showed is in\n[reference\u002Frelay.md](reference\u002Frelay.md#verified-behavior). The R2 and R3 runs passed the prompt\nas an argument; their stdin form below has not been re-probed.\n\n| # | From | To | Command shape | Status |\n|---|---|---|---|---|\n| R0 | any | same lane | Built-in `ListAgents` \u002F `SendMessage`. Not this skill | n\u002Fa |\n| R1 | WSL | this machine, Windows | `cd \u003Cwin-dir> && \u003Cclaude-exe> \u003Cagent-args> \"\u003Cprompt>\" \u003C \u002Fdev\u002Fnull` | tested: script (`powershell.exe`), list, multi-turn, message a `-n` receiver |\n| R2 | WSL | other machine, WSL | `printf '%s\\n' '\u003Cprompt>' \\| \u003Cssh> \u003Cwsl-alias> 'claude \u003Cagent-args>'` | tested: script, multi-turn, stream-json, `--bg`, message an interactive session and a `-p` receiver across accounts |\n| R3 | WSL | other machine, Windows | `printf '%s\\n' '\u003Cprompt>' \\| \u003Cssh> \u003Cwsl-alias> 'cd \u003Cwin-dir> && \u003Cclaude-exe> \u003Cagent-args>'` | reaches `claude.exe`; needs `\u002Flogin` at that machine's console |\n| R4 | Windows | this machine, WSL | `wsl.exe -d \u003Cdistro> --cd \u002Ftmp --exec \u003Cshell> -lc 'claude \u003Cagent-args> \"\u003Cprompt>\" \u003C \u002Fdev\u002Fnull'` | tested: script from a native-Windows `claude.exe` (its Bash tool, Git Bash); list |\n| R5 | Windows | other machine, WSL | `printf '%s\\n' '\u003Cprompt>' \\| ssh.exe \u003Cwsl-alias> 'claude \u003Cagent-args>'` | untested from a Windows origin (same hop as R2) |\n| R6 | Windows | other machine, Windows | `printf '%s\\n' '\u003Cprompt>' \\| ssh.exe \u003Cwsl-alias> 'cd \u003Cwin-dir> && \u003Cclaude-exe> \u003Cagent-args>'` | untested from a Windows origin (same hop as R3) |\n\nOver SSH a `-p` turn's prompt goes on stdin, and the quoted remote command carries only the flags,\nso no shell on either host parses prompt text. Never add `ssh -n` to an SSH row: it replaces stdin\nwith `\u002Fdev\u002Fnull` and drops the prompt. The local hops (R1, R4) keep the prompt as the last\nargument. A verb with no `\u003Cprompt>` drops the `printf` on an SSH row and the quoted prompt on R1\nand R4. `--bg` is not `-p` and takes its prompt as an argument, so a background session's prompt\nstays inside the remote command. In the tables here and in FLEET.md, `\\|` is an escaped `|`: type a bare `|`.\n\nA script with no agent goes over the same hop with the command in place of the `claude` part; on\nR1 that is `powershell.exe -NoProfile -Command '\u003Ccmd>'` from a Windows-side directory. A Windows\nscript on another machine goes over port 22 instead: `\u003Cssh> \u003Cwin-alias> '\u003Cpwsh command>'` (see\n[Port 22](#port-22-is-not-an-agent-lane)).\n\nPlaceholders, each filled from FLEET.md or a probe, never guessed:\n\n- `\u003Cssh>`: the Windows OpenSSH client. `ssh.exe` from Windows; from WSL,\n  `\u002Fmnt\u002Fc\u002FWindows\u002FSystem32\u002FOpenSSH\u002Fssh.exe`, because the distro holds no key.\n- `\u003Cwsl-alias>` \u002F `\u003Cwin-alias>`: the target's `wsl-shell` (port 2222) and `windows-shell` (port 22)\n  aliases.\n- `\u003Cclaude-exe>`: `\u002Fmnt\u002Fc\u002FUsers\u002F\u003Cuser>\u002F.local\u002Fbin\u002Fclaude.exe`, by absolute path.\n- `\u003Cwin-dir>`: a scratch directory under `\u002Fmnt\u002Fc\u002FUsers\u002F\u003Cuser>\u002F` (`claude.exe` needs a Windows-side\n  working directory), never a repository that defines the fleet's accounts or firewall.\n- Use FLEET.md's `\u002Fmnt\u002Fc` paths as given: never check them with file tools (Read, Glob, Grep),\n  and never ask a subagent's brief to. The hop check in\n  [Verify the hop](#verify-the-hop-before-you-trust-it) is what proves them.\n- `\u003Cdistro>`: from `wsl.exe -l -q`. `\u003Cshell>`: the distro account's login shell, so `claude` is on\n  `PATH`.\n\nPer-route detail, the receiver's nested quoting, reply patterns, background sessions, and the\nmechanisms not used here (Remote Control, cloud sessions, Channels) are in\n[reference\u002Frelay.md](reference\u002Frelay.md).\n\nConverging dotfiles on any fleet lane is not a route in this matrix: run `fleet-converge \u003Clane>`,\nby its bare name. It does not prompt because dotfiles renders an exact allow rule for that command\non every lane; no `Bash(ssh …)` rule is added, and the posture below stands for every other verb.\n\n- **Pointer**: for what `fleet-converge` checks and how it reaches each lane, read dotfiles\n  [docs\u002Fdesign-decisions.md, \"Converging a lane: `fleet-converge`\"](https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fdotfiles\u002Fblob\u002Fmain\u002Fdocs\u002Fdesign-decisions.md#converging-a-lane-fleet-converge).\n\n## Verify the hop before you trust it\n\nRun the row with `hostname &&` before the agent, `-p` as `\u003Cagent-args>` and `echo ok` as\n`\u003Cprompt>`, for example `printf '%s\\n' 'echo ok' | \u003Cssh> \u003Cwsl-alias> 'hostname && claude -p'`: the\nhostname proves which machine answered, `ok` that its Claude is signed in. A Windows lane answering with an\nexpired-OAuth-session authentication error routed correctly and is signed out: someone runs\n`\u002Flogin` at that machine's console (or over RDP). Nothing remote fixes it, and no credential\ntravels in a prompt.\n\n- **Pointer**: when matching a lane's answer to a signed-out state, fetch\n  [Login expired](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Ferrors#login-expired) live.\n\n## Verbs\n\nEach verb fills `\u003Cagent-args>` and `\u003Cprompt>` in a matrix row. Every verb but the first starts an\nagent, and each agent turn costs real usage, even a one-line turn\n([Manage costs: Track your costs](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcosts#track-your-costs); the Cost\nrow of [reference\u002Frelay.md § Verified behavior](reference\u002Frelay.md#verified-behavior) holds the\ndated measurement and how to re-measure it). When a script can do the job, run the script.\n\n| Verb | `\u003Cagent-args>` | `\u003Cprompt>` | Status |\n|---|---|---|---|\n| Run a script | none: the command itself replaces `claude ...` over the same hop | none | tested on R1, R2, R4; port 22 untested |\n| Prompt | `-p` | the task | tested on R1, R2, R4 |\n| Multi-turn | `-p --session-id \u003Cuuid>`, then `-p --resume \u003Cuuid>` against the same lane | each turn's prompt | tested on R1, R2 |\n| One open pipe | `-p --input-format stream-json --output-format stream-json --verbose` | none: user messages as NDJSON on stdin | tested on R2 |\n| List sessions | `-p` | `List the sessions you can reach` | tested on R1, R4 |\n| Message a session | `-p` | `SendMessage to \u003Cname>: \u003Ctext>` | tested on R1, R2 |\n| Named receiver | `-p -n \u003Cname> --settings '{\"crossSessionInbound\":\"accept\"}'` | the standing instructions | tested on R1, R2 |\n| Background session | `--bg --name \u003Cname> \"\u003Cprompt>\"`, not `-p`, with `\u003C \u002Fdev\u002Fnull` on an SSH row; manage with `claude agents --json --all`, `stop \u003Cid>`, `rm \u003Cid>` | none: it is in `\u003Cagent-args>` | tested on R2, trusted directory only |\n\n- Give every prompt the whole task: what done looks like, and what should make it stop and report.\n  Nobody answers a question a headless turn asks.\n- **Query and wait has no one-turn form.** Never wait on a `notify_when_idle` notice from a `-p`\n  sender: in our probe it never reached the sender\n  ([reference\u002Frelay.md § Verified behavior](reference\u002Frelay.md#verified-behavior)). Send the\n  message, then poll `claude agents --json` in that lane, or `--resume` the receiver or read its\n  output.\n  - **Pointer**: for how an idle notice is delivered and what a `-p` session can receive, fetch\n    [Get a notice when another session goes idle](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#get-a-notice-when-another-session-goes-idle)\n    and [Non-interactive sessions](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#non-interactive-sessions)\n    live.\n- **Multi-turn versus one open pipe.** The pipe keeps context in one process while the connection\n  stays open; per-turn `--resume` survives a disconnect. `--resume` ids belong to the lane that\n  made them, so resume against the same row. `--session-id` picks the id up front, so nothing\n  parses the first turn's output. Generate no id in this session: a command you hand over keeps\n  `\u003Cuuid>` for whoever runs it to fill, and a turn you run yourself can start with\n  `--output-format json` and resume with the `session_id` it returns (see \"Parse the JSON\" in\n  [Gotchas](#gotchas)). Say with the handed-over commands that the id exists only on that lane's\n  machine and account, so every follow-up goes to the same row.\n- **Background sessions** need a trusted working directory; anywhere else the start fails with a\n  workspace-not-trusted error and exit 1 (for its wording and fix, fetch\n  [Workspace not trusted when dispatching a background session](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Ferrors#workspace-not-trusted-when-dispatching-a-background-session)\n  live; as of 2026-10-11; recheck when that entry's message or fix changes). Address\n  `claude logs` and `claude attach` by the `id` from `claude agents --json --all` (bare `--json`\n  lists only active sessions); pass a name only after checking the lane's version against the CLI\n  reference. `logs` prints raw TUI output, so read a reply from the transcript (`--resume \u003Cid>`).\n  - **Pointer**: when deciding whether a lane can take a session name in place of the id, fetch\n    [CLI reference: CLI commands](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcli-reference#cli-commands) live.\n- A `-p` receiver needs `crossSessionInbound: accept`. For a session that must stay up, use a\n  background session, not a `-p` receiver. Give either one standing instructions as its prompt:\n  what to do with each message and when to stop. Delivery rules and the pointer are in\n  [reference\u002Frelay.md](reference\u002Frelay.md#a-named-receiver).\n  - **Pointer**: for when a `-p` run ends after its turn, fetch\n    [Run Claude Code programmatically: Background tasks at exit](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fheadless#background-tasks-at-exit)\n    live.\n- Use `--bare` for prompts, never for a receiver: a bare session cannot receive messages or be\n  listed.\n- Remote Control is interactive only: `-p --remote-control` does not connect. Keep it out of\n  headless recipes.\n- Session names belong to the target lane, and a listing shows only its own lane. List first, then\n  message a name from the listing that shows it. When a request names a machine and not a lane,\n  list each of that machine's lanes (on another machine, one R2 and one R3 turn).\n- **Pointer**: the receiver, `--bare` and Remote Control bullets rest on our probes and the flag\n  behavior in [reference\u002Frelay.md § Verified behavior](reference\u002Frelay.md#verified-behavior); when\n  one stops holding, fetch\n  [Control inbound messages](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#control-inbound-messages),\n  [Non-interactive sessions](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#non-interactive-sessions),\n  [Start faster with bare mode](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fheadless#start-faster-with-bare-mode)\n  and [CLI flags](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcli-reference#cli-flags) live.\n\n## Accounts are per lane\n\nEach lane signs into its own claude.ai account so one lane's usage never draws down another's (on\nthe desktop, WSL and Windows differ; the laptop's WSL lane uses a third). So:\n\n- **Same lane.** Sessions find each other through files and sockets; the peer tools work.\n- **Other lane, same machine.** WSL and Windows register under different homes and socket types;\n  each lane's `ListAgents` shows only its own lane.\n- **Other machine.** Remote Control lists only the signed-in account's sessions; across accounts\n  `ListAgents` shows nothing remote, even with Remote Control connected.\n\nThe peer tools are same-lane only, and no setting widens them; every other route starts a turn\ninside the target lane. Do not propose sharing an account: the split is the decision.\n\n- **Pointer**: the three facts are our probes' results\n  ([reference\u002Frelay.md § Verified behavior](reference\u002Frelay.md#verified-behavior)); when one is in\n  doubt, fetch\n  [See which sessions Claude can reach](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#see-which-sessions-claude-can-reach)\n  and\n  [Message sessions on other machines](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcross-session-messaging#message-sessions-on-other-machines)\n  live.\n\n## Port 22 is not an agent lane\n\n`ssh-admin` is a separate Windows account with no Claude sign-in and no DPAPI. Use port 22 for pwsh\nscripts only, with pwsh quoting; anything needing the console user's credentials goes through the\non-demand tasks FLEET.md lists.\n\n## Permission and safety posture\n\n- Every verb except running a script starts an agent in another lane, same machine included, and\n  these prompt under auto mode by design; let them prompt, and say so with every such command you\n  run, delegate or hand over. A script is judged on its own content.\n  When a prompt's cause is in doubt, fetch\n  [Eliminate prompts with auto mode](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fpermission-modes#eliminate-prompts-with-auto-mode)\n  live.\n- There is no `Bash(ssh ...)` allow rule anywhere in this fleet, and adding one is not the fix for a\n  prompt. If asked to stop the prompting, say what the rule would cost and decline.\n- Neither side runs `bypassPermissions`.\n- Never `wsl --shutdown` on a target, and never run `wsl -d` on a target's drift-convergence path:\n  both take the distro out from under whatever else uses it. The one exception is a start\n  `fleet-converge` makes itself, under the guard its pointer names; never run `wsl -d` on its\n  behalf.\n- On R1 and R4 the prompt travels in the command line, visible to the process table and shell\n  history; over SSH it travels on stdin, and the local shell history still records it. No secret\n  belongs in one.\n\n## Boundary\n\n| Neighbor | Owns |\n|---|---|\n| This skill | Every lane other than this session's: the other lane on this machine, and both lanes on other machines |\n| Built-in `ListAgents` \u002F `SendMessage` | Sessions in this lane, under this lane's account |\n| Built-in Remote Control | Human use of a lane's session from a phone or the web, under that lane's account. Not agent-to-agent across accounts |\n| `session-flow:continue-in-background` | A detached session in THIS lane |\n| `session-flow:orchestrate` | Delegation inside one session, to subagents |\n| `repo-fleet-hygiene:audit` | Fleets of REPOSITORIES. Same word, different subject |\n\n## Gotchas\n\n- **Use the Windows OpenSSH client.** It is agent-backed and is what `~\u002F.ssh\u002Fconfig` is wired to.\n  Git Bash's MSYS `ssh` reaches no agent and fails with `Permission denied (publickey)`, which reads\n  like a key problem and is not one.\n- **Give every `claude -p` turn a closed stdin.** Over SSH the piped prompt is stdin, and it ends\n  when `printf` does. On R1 and R4 redirect `\u003C \u002Fdev\u002Fnull`. A `-p` run reads stdin as its input, so\n  with no prompt argument the piped text is the prompt: see \"Pipe data through Claude\" and the\n  stdin-only example under \"Customize the system prompt\" in\n  [Run Claude Code programmatically](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fheadless#pipe-data-through-claude),\n  and the piped-content row of [CLI commands](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fcli-reference#cli-commands).\n- **Escape apostrophes in `\u003Cprompt>`.** Every SSH row single-quotes the prompt in\n  `printf '%s\\n' '\u003Cprompt>'`, and R4 single-quotes the whole command, on the LOCAL shell, so a `'`\n  in the prompt (\"what's\") closes that quote early. The rule follows the origin shell, not the\n  lane:\n  - **bash, zsh, and Git Bash** (a native-Windows Claude's Bash tool, R4 to R6): replace each `'`\n    with `'\\''`, or wrap the prompt argument (on R4, the command) in `$'...'` and write `\\'`. Doubling (`''`) silently drops the\n    apostrophe here: `'what''s'` arrives as `whats`.\n  - **PowerShell**, Windows PowerShell 5.1 or pwsh 7 (a native-Windows Claude's PowerShell tool,\n    or a PowerShell window): on R4, keep the prompt in doubled single quotes (`-p ''...''`) and\n    write an apostrophe as `''\\''''`. Never use `-p \"...\"` from PowerShell, with or without a\n    doubled `''` apostrophe. This form ran correctly on every version probed. Run the SSH rows\n    from bash, zsh or Git Bash, whose `printf` is a builtin.\n    - **Probe**: [PowerShell quoting to wsl.exe](https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fissues\u002F12908#issuecomment-6624493777).\n      **As of**: 2026-11-04. **Recheck trigger**: the\n      [about_Parsing](https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fpowershell\u002Fmodule\u002Fmicrosoft.powershell.core\u002Fabout\u002Fabout_parsing?view=powershell-7.6#passing-arguments-that-contain-quote-characters)\n      page changes how quotes pass to native commands.\n- **Use `wsl.exe --exec`, not `--`.** The quoting rules above assume `--exec`. For what each option\n  does, run `wsl.exe --help` on that machine, as\n  [Basic commands: Help command](https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fwindows\u002Fwsl\u002Fbasic-commands#help-command)\n  directs (the page has no section of its own for `--exec`).\n- **Probe interop by absolute path.** `cmd.exe \u002Fc` prints nothing inside an sshd session and reads\n  as \"interop is broken\" when it is not; run the `.exe` by absolute path.\n- **Parse the JSON, not the stream.** Treat stdout as able to carry lines that are not the JSON\n  result, and extract the object first, for example `grep '^{' | jq -r .session_id`. What a\n  session-end hook can print is in [Hooks: SessionEnd](https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fhooks#sessionend).\n- **A session name is not an address.** Hostname and lane, always.\n",{"metadata":24,"description":7,"argument-hint":27,"user-invocable":28,"disable-model-invocation":20},{"summary":25,"workflow-stage":26},"Reach another fleet lane (WSL or Windows, here or remote) to run, prompt, query or message","anytime","[relay]",true,"main","d0e68c3f36b92a1d6381f0ba0060d7a0b8b886c5","plugins\u002Ffleet\u002Fskills\u002Freach",7,"https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins","https:\u002F\u002Fgithub.com\u002Fmelodic-software","https:\u002F\u002Fmelodicsoftware.com","Independent software studio building developer tools, automation, and application services, with roots in music.","2026-11-15 05:15:06",{"source":8,"skillName":5,"commands":39,"prompt":49,"zipAvailable":28},[40,43,46],{"tool":41,"command":42},"npx","npx skills add melodic-software\u002Fclaude-code-plugins --skill reach",{"tool":44,"command":45},"bunx","bunx skills add melodic-software\u002Fclaude-code-plugins --skill reach",{"tool":47,"command":48},"pnpm","pnpm dlx skills add melodic-software\u002Fclaude-code-plugins --skill reach","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Ffleet\u002Fskills\u002Freach\u002FSKILL.md",[51],{"path":52,"resourceType":53,"resourceTypeDesc":54,"sizeBytes":55,"sourceHtmlUrl":56},"reference\u002Frelay.md","REFERENCE","参考文档",15216,"https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Ffleet\u002Fskills\u002Freach\u002Freference\u002Frelay.md",[58,65,72,79,84,91],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":19},"01M7DYYDG31B9YX7N9MB9QW2GW","track","Backlog reads through the bound tracker: search open and closed items, view one item with its comments and state history. Use when another skill searches or reads tracker items.","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fsession-flow\u002Fevals\u002Frecall-still-reports-reopened-bug\u002Fwork-items-stub\u002Fskills\u002Ftrack\u002FSKILL.md",[],[],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M7DYXVBAVG3NV30PA7GKW9Q6","reviewer-media","Reviewer media for one pull request: decides whether it needs a narrated walkthrough, builds the storyboard from \u002Freview:explain-change's digest, has \u002Fexplainer-video:produce render it and hands the manifest to \u002Fsource-control:pull-request media. Use when asked for 'reviewer media', 'a walkthrough video for this PR', 'a produced video of this change', or when pull-request ready or monitor hands a PR over.","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Freview\u002Fskills\u002Freviewer-media\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":19},"01M7DYRZ5YD44AMYH26YR25GZ8","route","Model and effort per agent role (orchestrator, worker, verifier, retrieval) for a multi-agent run, resolved through the bundled defaults and the user, team and personal override config layers into fenced JSON with single-agent and fan-out variants and the layer behind each value. Use when asked 'which model for the subagents', 'route the roles' or 'model routing', or before a skill passes a role's model on a spawn.","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fdiscovery\u002Fevals\u002Fex-model-pin-without-override\u002Fmulti-agent-stub\u002Fskills\u002Froute\u002FSKILL.md",[],[],{"skillNo":80,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M7DYRTVT4FWTGDJY0HCCFJMZ","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fdiscovery\u002Fevals\u002Fex-model-override-from-route\u002Fmulti-agent-stub\u002Fskills\u002Froute\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":19},"01M7DYNZC8PNW3T27V2ADKECKJ","principles","Module-design vocabulary for architecture findings, and the checks built on it. Use when: 'what is a deep module', 'what is a seam', 'deep vs shallow module', 'what is the deletion test', 'leverage and locality', 'module design vocabulary', 'which word do we use for X in architecture findings'. Scanning a codebase for friction is \u002Farchitecture:improve.","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Farchitecture\u002Fskills\u002Fprinciples\u002FSKILL.md",[],[],{"skillNo":92,"name":93,"nameZh":6,"description":94,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":19},"01M7DYNMCZNEWBFHV9KTSS6H60","write","Structured five-field bug report (title, steps to reproduce, expected vs actual, justified severity, suggested fix location) from an informal description, read-only. Use when the user names a defect they observed, or asks for one written up ('report a bug', 'file a bug', 'write this up as a bug').","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Farchitecture\u002Fevals\u002Fdeepening-defect-goes-to-bugs-write\u002Fbugs-stub\u002Fskills\u002Fwrite\u002FSKILL.md",[],[]]