Overview

octri is the command line for your Octri project. It does what the dashboard does: push a spec, build SDKs, publish docs, upload symbols, and work through the errors your SDKs report from production.

Install

npm install -g @octri/cli
octri --help

Node 20 or newer.

Sign in

octri auth login
octri projects use

auth login prompts for your email and password, and for your authenticator code if you have two-factor on. --email and --password work for scripted setups. projects use with no argument opens a picker; pass a project id to skip it.

The session lives in ~/.octri/config.json, mode 0600. Profiles keep separate environments side by side:

octri config use prod --api-url api.octri.dev
octri config set defaultLanguages go,rust,swift

What it covers

Run octri <group> --help for one group on its own.

A first run

octri specs push ./openapi.yaml
octri sdk build --lang go,typescript --download
octri docs versions publish <specId>

specs push waits for the whole pipeline, not the upload: the spec parses in a second, then the doc pages, search index and changelog are written behind it. The spinner names the stage it is on.

Output

Three modes, chosen automatically.

--json is the one to script against. NO_COLOR and --plain are honoured.

Global flags

Values resolve flag first, then environment variable, then the stored profile. The environment variables are OCTRI_PROFILE, OCTRI_API_URL, OCTRI_TOKEN, OCTRI_API_KEY, OCTRI_PROJECT_ID and OCTRI_CONFIG_DIR.

Monitoring

Reads and triage use your session, so they need nothing but a login:

octri monitoring summary
octri monitoring issues --status unresolved
octri monitoring issue <id>
octri monitoring resolve <id>

The two upload commands work differently. They talk to the monitoring service directly with an ingest token, because a CI job has one secret and no way to log in.

octri monitoring sourcemaps upload ./dist
octri monitoring sources upload ./src

Signed in, the CLI resolves the connection for you. In CI you pass it. octri monitoring config prints the three values to store as secrets.

See Source maps, Source files and CI setup, or Monitoring for the full command set.

Driving it from an AI assistant

octri mcp serve

That speaks MCP over stdio and exposes the same surface as typed tools, so an assistant can push a spec, wait for a build, read the generated source, and read the production errors that build caused.

{
  "mcpServers": {
    "octri": {
      "command": "octri",
      "args": ["mcp", "serve"]
    }
  }
}

Credentials come from your stored profile. The assistant never sees them. Reads are open; publishing to package registries and deleting a spec are refused unless you start the server with --allow-publish or --allow-delete.