What is the Orbyt CLI?

@orbyt/cli is the official terminal client for Orbyt Intelligence. It wraps the @orbyt/intelligence TypeScript SDK in a TTY-friendly command-line interface, so engineers, data teams, and CI/CD pipelines can hit the six AI engines without writing code. Commands include calculate, history, projections, compare, search, demo, lineage, and skills.

npm install -g @orbyt/cli installs globally. Then run orbyt login to paste your API key. Keys are stored in a local config file at ~/.orbyt/config.json and re-used across commands. Homebrew tap and curl-install instructions are documented on the page.

Does the CLI support JSON output?

Yes. Every command accepts --json to produce machine-readable output suitable for piping to jq, scripting, or consuming from a CI/CD pipeline. Without --json, output is pretty-printed for human reading with colors, units, and contextual notes.

Can I use it in CI/CD?

Yes. Set the ORBYT_API_KEY env var (or pass --api-key=...) and the CLI runs without needing a login config. All commands return appropriate exit codes (0 for success, 1 for API errors, 2 for usage errors) so CI scripts can branch on the result.

Does it tunnel webhooks for local development?

Yes. orbyt webhooks listen starts a local tunnel and forwards webhook deliveries from your Orbyt account to localhost. The tunnel keeps the connection open, retries on disconnect, and prints each delivery with its signature header so you can verify locally.

Is there an MCP server bundled?

Yes. orbyt mcp serve runs the @orbyt/mcp stdio transport against your API key. Drop this into Claude Code's mcp.json or any agent's MCP config to grant the agent access to the six locked tools (analyze_compensation, analyze_skills, analyze_market, discover_roles_and_cities, find_adjacent_opportunities, list_capabilities).

@orbyt/cli · macOS · Linux · Windows · From $99/mo on Build

Orbyt Intelligence CLI.

Terminal-first access to six AI engines. TTY-friendly output, JSON mode for scripts, MCP stdio bundled, webhook tunneling, full parity with the @orbyt/intelligence SDK.

API docs MCP integration

Install.

Three install paths. Pick the one that fits your environment.

npm (recommended)

npm install -g @orbyt/cli

Homebrew (macOS / Linux)

brew tap orbytjobs/cli
brew install orbyt

curl install (any POSIX)

curl -fsSL https://orbytjobs.ai/install.sh | sh

Authentication and config.

Run orbyt login and paste your API key when prompted. Keys are stored at ~/.orbyt/config.json with mode 0600 (owner-only read/write). The key is re-used across every command in the session.

$ orbyt login
✓ Paste your API key (starts with intelligence_live_ or intelligence_test_):
✓ Key validated. Tier: pro. Scopes: intelligence:read, skills:read, market:read.
✓ Saved to ~/.orbyt/config.json (0600)

In CI/CD, skip orbyt login and pass the key via env var or flag.

# CI/CD env var
ORBYT_API_KEY=intelligence_live_... orbyt salaries calculate --role=ai-engineer --city=san-francisco --json

# Explicit flag
orbyt --api-key=intelligence_live_... salaries calculate --role=ai-engineer --json

Restricted API keys with per-engine scopes are supported. A key with only intelligence:read can hit orbyt salaries calculate but not orbyt salaries skills (which needs skills:read).

Common workflows.

Calculate a single salary

orbyt salaries calculate \
  --role=ai-engineer \
  --city=san-francisco \
  --level=senior \
  --company-stage=series-b

Pull skill premiums for a role

orbyt salaries skills \
  --role=ai-engineer \
  --city=san-francisco \
  --limit=20 --json | jq '.data[] | { skill, premium, ci_low, ci_high }'

Project comp 4 years forward

orbyt salaries projections \
  --role=llm-systems-engineer \
  --city=san-francisco \
  --through=2030

Compare two roles in the same city

orbyt salaries compare \
  --role-a=ai-engineer \
  --role-b=ml-engineer \
  --city=san-francisco

Search the role catalog (with confidence scores)

orbyt salaries search "llm platform engineer"

# Returns top-3 matches:
#  1. role:llm-platform-engineer        (confidence 0.97)
#  2. role:llm-systems-engineer         (confidence 0.78)
#  3. role:ml-platform-engineer         (confidence 0.62)

Resolve a data point's full provenance

orbyt lineage \
  ai-engineer:san-francisco:median_base:2026-Q2

# Prints the full source trail: BLS OES record, H-1B LCA filings,
# job postings, community submissions, and the reconciliation rules.

MCP bundled.

Run orbyt mcp serve to start the MCP server over stdio, using your CLI's authentication. Drop this into Claude Code's mcp.jsonor any agent's MCP config to grant the agent access to the six locked Decision-Ready tools.

// .claude/mcp.json
{
  "mcpServers": {
    "orbyt-intelligence": {
      "command": "orbyt",
      "args": ["mcp", "serve"],
      "env": {
        "ORBYT_API_KEY": "intelligence_live_..."
      }
    }
  }
}

The CLI keeps the stdio process running, handles request cancellation and reconnection, and reports telemetry to a local log file at ~/.orbyt/mcp.log. See the MCP integration page for the full tool catalog and Decision-Ready Response shape.

Webhook tunneling for local dev.

When testing webhook handlers locally, run orbyt webhooks listen --forward-to=http://localhost:3000/api/webhooks/orbyt and the CLI opens a secure tunnel that forwards every webhook delivery from your Orbyt account to your local server. Signature headers and event payloads come through unchanged so signature verification works end-to-end.

$ orbyt webhooks listen --forward-to=http://localhost:3000/api/webhooks/orbyt
✓ Tunnel established: https://tunnel.orbytjobs.ai/u/a4f9b21c
✓ Forwarding to http://localhost:3000/api/webhooks/orbyt
✓ Ctrl-C to disconnect

[2026-05-11T18:32:14Z] data.refreshed (req_a4f9b21c8e7d3f01)  →  200 OK (42ms)
[2026-05-11T18:35:02Z] data.corrected (req_b5g0c32d9f8e4g12) →  200 OK (28ms)
[2026-05-11T18:37:55Z] data.refreshed (req_c6h1d43e0g9f5h23) →  200 OK (39ms)

Signatures are computed against your Orbyt webhook signing secret and forwarded unchanged. Run orbyt webhooks verify [event-id]to re-verify a specific delivery's signature after the fact.

Bulk operations.

For exports, batch lookups, and CSV ingestion, use orbyt salaries export. The command paginates the API, writes CSV or JSON Lines, and respects rate limits (auto-throttles when X-RateLimit-Remaining approaches zero).

# Export every salary point for AI roles across every tracked U.S. city
orbyt salaries export \
  --role-family=llm-and-agents \
  --all-cities \
  --output=ai_salaries_2026Q2.csv

# Batch lookup from an input CSV
orbyt salaries calculate \
  --batch-input=requests.csv \
  --output=results.csv \
  --concurrency=4

Rate-limit handling is transparent. The CLI watches X-RateLimit-Remaining and X-RateLimit-Reset headers and auto-throttles. Failed requests retry up to 3 times with exponential backoff. Requests that ultimately fail are written to --failed-output=failures.csv so you can re-run them later.

CI/CD integration.

The CLI is designed for CI scripts. Exit codes are stable: 0 on success, 1 on API errors, 2 on usage errors, 3 on auth errors, 4 on rate-limit errors that exhausted retries. Pipelines can branch on the result.

# .github/workflows/refresh-comp-bands.yml
- name: Refresh AI engineer comp bands
  env:
    ORBYT_API_KEY: ${{ secrets.ORBYT_API_KEY }}
  run: |
    orbyt salaries export \
      --role=ai-engineer \
      --cities=san-francisco,new-york,seattle,austin \
      --output=bands.csv \
      || (echo "::error::Orbyt export failed" && exit 1)
    git add bands.csv
    git commit -m "chore: refresh AI comp bands ($(date +%Y-%m-%d))"

For long-running jobs, use orbyt jobs submit and orbyt jobs status to offload export work to the server-side bulk endpoint. Available on the Scale tier.

Reference.

The CLI's auto-generated reference is available at orbyt --help and orbyt <command> --help. Every command lists its arguments, flags, examples, and required scopes. The reference is also shipped as part of the npm package's README.

Source code at github.com/orbytjobs/cli. Issues at the same repo. Changelog at orbyt version --changelog.

See also.

API docs

Canonical API reference

MCP integration

Six tools, one endpoint

Webhooks

Event types and signature verification

All six engines

The full engine catalog

Last updated May 2026. Package: @orbyt/cli on npm. Documentation auto-generated from the SDK's OpenAPI spec.