npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@runtime-judgement/cli

v0.1.0

Published

Runtime Judgement CLI — same ops as the UI, scriptable. Implements P8 (programmable in both directions) from the AI-native principles ADR.

Vulnerabilities

Powered byPowered by Snyk
  • 0High
  • 0Medium
  • 0Low

Links

Maintainers

rossamac01rossamac01

Keywords

runtime-judgementclimcpagentsdebugging

Readme

@runtime-judgement/cli

The rj command-line interface for Runtime Judgement — the failure-attribution debugger for multi-agent AI systems.

P8 of the AI-native principles ADR (dev-docs/strategy/2026-05-26-adr-ai-native-principles.md in the main repo) says: if you can click it, you can curl it. This package is the CLI half of that contract — everything available in the agent-page UI is also available as rj <noun> <verb>.

This first cut ships 6 read commands (agents list, tests list, runs get, headline get, whoami, login). Write commands (mutating snapshots, runs, pipeline) land in a follow-up release alongside the MCP write tools.

Quickstart

# 1. Install (Node 18+)
npm install -g @runtime-judgement/cli

# 2. Authenticate — opens your browser, mints a key, writes ~/.rj/token
rj login

# 3. Verify
rj whoami

That's it. From here, rj agents list, rj tests list <agent-id>, etc.

Other installers

pnpm add -g @runtime-judgement/cli      # pnpm
yarn global add @runtime-judgement/cli  # yarn classic

# No global install — one-shot via npx:
npx -y @runtime-judgement/cli@latest whoami

# From source (contributors):
git clone https://github.com/proveai/runtime-judgement-app
cd runtime-judgement-app
pnpm install
pnpm --filter @runtime-judgement/cli build
node packages/rj-cli/dist/index.js --help

Configure

Recommended: rj login.

rj login

The CLI opens your browser to the settings page, which mints a fresh API key and posts it back to a one-shot local listener. The token is written to ~/.rj/token (0600) and you're ready to go. No copy/paste required.

For headless environments (SSH, CI bootstrap, no browser), use --paste:

rj login --paste
# Prompts: "Paste your rj token (input hidden): "
# Generate the token at <base-url>/app/settings/api-keys first.

Other flags:

  • --base-url <url> — override the deployment URL (default: https://runtime-judgement-app.vercel.app).
  • --no-open — print the login URL instead of auto-opening it. Useful for SSH or remote desktops where the local browser isn't the right one.
  • --timeout-ms <ms> — how long to wait for the browser callback (default 300000 / 5 min).

Manual configuration

If you'd rather skip rj login and manage credentials yourself, the CLI resolves config in this order:

  1. RJ_BASE_URL env var (default: https://runtime-judgement-app.vercel.app)
  2. RJ_AUTH_TOKEN env var, OR
  3. The contents of ~/.rj/token (one line, no quotes)

Generate a token at <base-url>/app/settings/api-keys.

Quick check:

export RJ_AUTH_TOKEN=rj_live_…
rj whoami

Expected output:

rj
  base URL   https://runtime-judgement-app.vercel.app
  token      rj_live_abcd…wxyz
  OK

Commands

rj agents list

List the agents you can see. Default output is a table; pass --json for raw output.

rj agents list
rj agents list --limit 5
rj agents list --json | jq '.agents[].id'
ID                       NAME             LAST RUN              HEALTH
-----------------------  ---------------  --------------------  ----------
agt_demo_cx_support      cx-support       2026-05-26T17:42:00Z  regressed
agt_demo_payments        payments-agent   2026-05-26T16:11:09Z  green

rj tests list <agent-id>

List tests (snapshots) attached to one agent. --filter is a substring match on test name.

rj tests list agt_demo_cx_support
rj tests list agt_demo_cx_support --filter "order"
rj tests list agt_demo_cx_support --json

rj runs get <run-id>

Fetch the full per-test breakdown for a single run.

rj runs get run_01HZ123…
rj runs get run_01HZ123… --json | jq '.perTest[] | select(.verdict != "PASSED")'

rj headline get <agent-id> [--section]

Read the judge-authored P1 headline for an agent section. Defaults to overview; other sections are tests, suites, runs, pipeline, settings.

rj headline get agt_demo_cx_support
rj headline get agt_demo_cx_support --section pipeline
rj headline get agt_demo_cx_support --json

rj whoami

Diagnostic — prints the resolved config (with the token truncated for safety) and runs a connectivity check against agents.list.

rj whoami

rj login

Authenticate the CLI. By default this opens your browser to the settings page, mints a key, and writes it to ~/.rj/token. See Configure above for --paste, --no-open, --base-url, and --timeout-ms.

rj login
rj login --paste                   # headless / SSH
rj login --no-open                 # print URL instead of opening
rj login --base-url https://staging.runtimejudgement.com

Sample output (browser flow):

rj login
  base URL   https://runtime-judgement-app.vercel.app
  callback listener on http://127.0.0.1:57563
  opening https://runtime-judgement-app.vercel.app/app/settings/api-keys?cli=1&port=57563
  waiting for browser…
  wrote token to /Users/you/.rj/token
  verifying…
rj
  base URL   https://runtime-judgement-app.vercel.app
  token      rj_live_abcd…wxyz
  OK
rj login: success.

Troubleshooting

rj: no auth token found. You haven't authenticated yet. Run rj login. For headless boxes (CI, SSH, no browser) use rj login --paste and paste a token minted from <base-url>/app/settings/api-keys.

rj login opens the wrong browser (remote desktop, SSH-with-X-forwarding). Use rj login --no-open to print the URL and visit it manually, or rj login --paste to skip the browser entirely.

rj whoami fails with a 401 / "unauthorized". Your token is invalid, revoked, or pointed at the wrong deployment. Re-mint at <base-url>/app/settings/api-keys, then rj login --paste (or overwrite ~/.rj/token). If you've recently moved deployments, also unset / reset RJ_BASE_URL.

rj whoami fails with a network error. Confirm the base URL is reachable from your shell: curl -fsS $(rj whoami --json | jq -r .baseUrl)/api/health should return 200. If you're behind a corporate proxy, ensure Node's HTTPS_PROXY env var is set.

I want to point the CLI at a staging / self-hosted deployment. Set RJ_BASE_URL=https://<your-host> (no trailing slash). rj login --base-url <url> does the same for one-off login flows.

Where's ~/.rj/token stored? ~/.rj/token is a one-line file (0600 permissions). Delete it to log out. Multi-account support is not in scope today — use RJ_AUTH_TOKEN env vars to switch on a per-command basis.

Architecture

  • HTTP JSON-RPC to /api/mcp on the host deployment.
  • One module per command under src/commands/. Each command exports both a commander.Command and a run…() function so tests can call the logic without parsing argv.
  • src/mcp-client.ts is a tiny custom client. We do not depend on @modelcontextprotocol/sdk because we POST a single envelope and parse the response synchronously — the SDK is geared toward stdio/SSE.

Tests (contributors)

pnpm --filter @runtime-judgement/cli test

The suite mocks fetch end-to-end; no network calls are made.

License

MIT.