Operate the `sno` command-line tool: install, update, check, and uninstall Sno products, read the exact commands and flags of the installed version, and run the account, skills, onboarding, Station telemetry and REM commands. Use when the owner or a task says "sno", "Sno CLI", "sno setup", "sno update", "sno doctor", "sno uninstall", "sno station", "telemetry consent", "rem verdict", "sno account", or "sno skills". The exact flags always come from the installed binary, never from this file. Do not use it to decide the owner's choices for them.
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
sno is one binary. This file carries the rules, the task map and the recipes. The exact
commands and flags of the version that is installed are served by the binary itself, so they
cannot be out of date.
Choose the executable once and keep using it:
type -a sno (works in bash and zsh). If two sno files are listed, or the version is older
than the installer said, tell the owner, use the newer absolute path, and note that a
running shell may need hash -r or a new terminal.sno is not installed, say so. Do not look for its source.In every command block, sno stands for that absolute executable. Then read the reference
for this exact version, once:
sno --version
sno
sno alone prints the list of commands. Before you use a command, read sno <command> --help
(for a nested command, sno <group> <command> --help). That output comes from the installed
binary, so it is the authority for every flag.
--json when you parse the output. Stdout is then exactly one JSON value.notes in JSON, the first line in text)."usage": true and
"executed": false, or full help plus an example, and the exit code is 0. A normal result has
no executed field; a missing executed means the command ran.sno station consent
with a misspelled level only shows the current level and a note, and the exit code is still 0.
Check that the returned value is the one the owner chose, and that executed is not false.When a Sno command invokes Sno GPU, keep the real project/session/run context in the
command's supported inputs; read its installed help before choosing flags or JSON fields.
For RSI, sno rem judge already accepts a run with per-session project context and
sno rem recall accepts project_id. Do not relabel a mixed-project run as the shell's
current repository, invent attribution flags, or call GPU/Helicone directly to bypass Sno.
The component sending GPU HTTP must supply its own repository slug as X-Sno-Source, the
stable feature workload (rsi for RSI) as X-Sno-Workload, and a fresh UUID v4 as
X-Request-ID generated inside every attempt, including retries, never in static client
headers. New repositories and workloads need no registration or approval. The sender must
record status or failure with that ID and preserve existing operation IDs. Authentication/admission supplies verified user
and machine identities; never add caller-supplied user/machine identity headers. Missing
diagnostic context is recorded as unknown without rejecting valid model work. These rules
also apply to future GPU-backed commands; the full caller contract is in the skill library's
skills/README.md. Other model providers and health/model-list probes keep their own behavior.
| Task | Command |
|---|---|
| Install the default product, or a chosen one | sno setup, sno setup <product> |
| Update the CLI and installed products | nothing to run: any sno command updates first when a newer release is out |
| Health of programs, skills and Station | sno doctor |
| Project memory and improvement records | sno project status, sno project list |
| Remaining model allowance | sno usage |
| Uninstall | sno uninstall, sno uninstall <product>, sno uninstall all |
| Machine identity | sno account register, sno account claim |
| Agent instructions served by the binary | sno skills, sno skills <name> |
| Reply to an optional product offer | sno products |
| Configure and prove Station | sno onboarding status, sno onboarding apply, sno onboarding verify |
| Telemetry choices and export | sno station consent, pause, resume, export |
| Verify a stored event | sno station audit |
| Local REM jobs | sno station rem-start, sno station rem-status |
| REM runs and verdicts | sno rem judge, sno rem recall, sno rem verdict |
Flags are not listed here on purpose; read them from sno <command> --help.
sno --version
sno doctor --json
sno skills core
Tell the owner the version, where it is installed, and the one next step. Station onboarding questions and product offers are put to the owner as questions; you never answer them.
sno doctor --json
sno project status --json
sno usage --json
doctor exits 0 on warnings. Read the rows whose result is failed or missing and relay
each with its fix. If usage says the machine is not registered, tell the owner and offer
sno account register.
Never ask the owner to run sno update. Any sno command that finds the CLI or an installed
product older than the latest release (checked once a day) updates first and then runs what was
asked. sno update exists for manual use only.
sno uninstall
sno uninstall sno-station
sno uninstall sno-station --yes
The first two only list and explain; nothing is removed. Uninstalling is the one action that is
asked about even when the owner just told you to do it, the same way rm -rf is: a request such
as "uninstall sno-station" is not the confirmation. Run the first two, show the owner what will be
removed, ask "Uninstall now?", and stop. Run the third only after the owner answers yes in a
later message. If there is no later message, your answer is the plan and the question.
--purge-state also deletes Reach mail state; ask for that separately.
sno station rem-start --type noop --scope persona:demo
sno station rem-status --wait --timeout 60
rem-status without an id reads the most recent job started on this machine.
Ask the owner, and wait for the answer, before you:
sno station consent <level>) or pause or resume it;sno products <product> <answer>);sno rem verdict);sno uninstall … --yes, or --purge-state — even when the owner's own request was to uninstall.Reading is always fine: sno station consent with no value, sno products with no arguments (it lists), doctor,
usage, skills, project.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Answer how long agent work will take, how much longer active work needs, or what clock time applies. Estimate full AI-agent wall time, measure active progress first, and get owner-facing times from `sno report-time`. Never estimate human-developer time.
The English summary is being prepared. Showing the original description.
Prove that a reported run caused its intended real effect. Walk the executor path from entry to effect, capture one observable per hop, and report a result a no-op could not produce.
The English summary is being prepared. Showing the original description.
Give the owner one page when they come back from sleep, a meeting or a long run: what got done, what is stuck, what waits for their answer, and what it cost in quota. Use when the owner asks what happened while they were away, wants a status page, or when you finish a long stretch of work and must report it.
The English summary is being prepared. Showing the original description.
Report how many problems the two brains found: mutual review (the other model checks the work, from the peer-review records) and self-check (an agent finds its own error in its own conversation, judged by that agent's own model), side by side. Use when the owner asks whether the second agent or the reviews are worth it, or wants the week's review results.
The English summary is being prepared. Showing the original description.
Write, revise or review a charter: the one document that tells an executor what to deliver, what is out of bounds, what the owner already decided, and how each success check will be proven. Use when work needs a written brief before someone else does it, or when asked to write, change or check a charter.
The English summary is being prepared. Showing the original description.
MANUAL-ONLY: load solely when the human owner types `/cos`, uses `$cos`, or asks you to act as the Chief of Staff (COS). The COS is the one agent the owner talks to; it coordinates the owner and the per-project Project Leads (PLs). A task card, charter, or dispatch makes you an executor, not COS.
The English summary is being prepared. Showing the original description.