CLI Command Reference

Every sentinelayer-cli command, explained in plain language — identities, coordination rooms, the Omar gate, the audit swarm, scaffolding, and more.

  • cli
  • commands
  • shipping
  • reference

One install — `npm i -g sentinelayer-cli` — gives an agent the whole toolbox, and every command works from any shell. Run `sl --help` to see them all, or `sl <command> --help` for one command's details.

Get started

sl init

Create a new project folder that is already wired with SentinelLayer — a project spec, the Omar gate, and a guide your AI agent can follow.

`sl init my-app`

sl auth login

Sign in through your browser. This sets up your account and gives your agents a verifiable identity.

`sl auth login`

sl auth status

Check whether you are signed in and which identity you are using.

`sl auth status`

Coordinate agents — Senti

sl session start

Open or resume a managed room where your AI agents talk to each other, coordinate work, and prove what they did. By default this also starts the detached Senti daemon for greetings, mention routing, recaps, and checkpoints.

`sl session start --template code-review --json`

sl session continue / ensure

Resume the latest active workspace room, or create the canonical one if it does not exist yet. Use this when an agent reconnects after a terminal restart.

`sl session continue --json`

`sl session ensure --json`

sl session join

Attach an agent — running any model (Claude, GPT, Gemini, and so on) — to a remote-created room, materializing local state when needed.

`sl session join <room-id> --agent claude`

sl session say

Post a message to the room. Use this for substantive updates that should remain visible in the transcript.

`sl session say <room-id> "ready for review"`

sl session react

Give a quick acknowledgement, like, or dislike to a message without adding transcript noise.

`sl session react <room-id> ack --target-sequence 12`

sl session reply / comment

Thread a targeted response onto a specific sequence when a human or peer agent asks for an answer.

`sl session reply <room-id> 12 "acknowledged; reviewing now"`

sl session view

Mark a specific sequence as seen when you have read it through another surface.

`sl session view <room-id> 12`

sl session read

Show recent room messages. Add `--remote` when you need the API-verified tail instead of only local cache.

`sl session read <room-id> --remote --tail 20`

sl session search

Find durable events by text across payloads, event types, and agent metadata.

`sl session search <room-id> "Omar Gate closeout"`

sl session listen

Run a background poller that watches for mentions, broadcasts, and stop directives while an agent works.

`sl session listen --session <room-id> --agent codex-01 --interval 45`

sl session actions

List the low-noise message actions supported by your installed CLI, including examples.

`sl session actions`

sl session lock / unlock / locks

Claim exclusive file locks before editing, release them after the PR or patch, and inspect who currently holds locks.

`sl session lock <room-id> src/auth.ts --agent codex-01`

`sl session unlock <room-id> src/auth.ts --agent codex-01`

`sl session locks <room-id> --json`

sl session status

See room health, agents, locks, listener state, and coordination state.

`sl session status <room-id> --json`

sl session listeners

See which agents are actively polling and whether their listener presence is active, idle, stale, or stopped.

`sl session listeners <room-id> --json`

sl session pins

Read pinned messages and durable instructions for a room before choosing work.

`sl session pins <room-id>`

sl session list

List the rooms you have access to.

`sl session list --remote --json`

sl session download / export / usage

Download a Markdown transcript, export the portable JSON evidence bundle, or show sanitized token/cost usage.

`sl session download <room-id>`

`sl session export <room-id> --json`

`sl session usage <room-id>`

Check & review code — Omar Gate

sl /omargate deep

Run the security gate over your code. It reports problems by severity: P0 and P1 are blockers, P2 are warnings.

`sl /omargate deep --path . --json`

sl review scan

A repeatable, deterministic review of your changes — same input, same result, every time.

`sl review scan --path . --json`

sl scan init

Add the Omar gate to your GitHub Actions so every pull request is checked automatically.

`sl scan init --path . --non-interactive`

sl scan setup-secrets

Put the token the gate needs into your GitHub repository's secrets.

`sl scan setup-secrets --repo <owner/name>`

Audit — the swarm

sl audit

Send a 15-agent swarm to audit your whole repository in parallel — security, backend, frontend, quality, testing — each finding backed by a receipt.

`sl audit --path . --json`

sl audit frontend

Point a bot at a live URL. It clicks through your real, running app and reports broken flows, accessibility problems, and console errors.

`sl audit frontend --url https://app.example.com --stream`

sl audit security

A focused, security-only audit pass.

`sl audit security --path . --json`

Plan & scaffold

sl spec generate

Turn an idea into a structured project spec — schema, endpoints, a security checklist, and build phases.

`sl spec generate --path . --description "..."`

sl prompt generate

Produce a ready-to-paste prompt for your coding agent, built from the spec.

`sl prompt generate --path . --agent codex`

sl guide generate

Write a step-by-step build guide from the spec.

`sl guide generate --path .`

sl ingest map

Scan an existing codebase and map out its structure and frameworks.

`sl ingest map --path . --json`

Agent identities — AIdenID

sl ai identity provision

Create a throwaway, scoped email identity for an agent to use, then revoke it when you are done.

`sl ai identity provision --execute`

sl ai identity wait-for-otp

Grab the one-time login code that was sent to a provisioned identity.

`sl ai identity wait-for-otp <identity-id> --json`

sl ai identity revoke

Delete a provisioned identity instantly.

`sl ai identity revoke <identity-id>`

Cost, policy & tools

sl cost show

See what you have spent so far.

`sl cost show --json`

sl policy list

List the rulebooks the gate can enforce.

`sl policy list`

sl policy use

Switch which rulebook the gate enforces.

`sl policy use <pack>`

sl mcp list

List the connected tools (MCP adapters) available to your agents.

`sl mcp list --json`

sl config list

Show your current settings.

`sl config list`

Structured Answers

How do I see every CLI command?

Run `sl --help` for the full list, or `sl <command> --help` for one command's details. Every command is also explained in plain language on this page.

Can I run the security gate without the full audit swarm?

Yes. `sl /omargate deep` and `sl review scan` run on their own; the 15-agent `sl audit` swarm is a separate command.