CLI

pip install litmus-ai-sdk installs the litmus-ai command. It reads LITMUS_URL, LITMUS_TOKEN and LITMUS_ORG, or the file litmus-ai init writes. Every command below is a client of the routes in this reference: the routes listed are the ones its code can reach, found from the CLI’s own call graph. A command with none listed calls no route this reader can see; most of those work on your machine only.

Command	Routes it calls
`litmus-ai init`	`GET /v0/whoami` 1 call(s) that need the workspace’s service token, not an org token
`litmus-ai probe`	none found
`litmus-ai call`	none found
`litmus-ai record`	none found
`litmus-ai push`	`GET /v0/evidence` `PATCH /v0/orgs/{orgtag}/sources/{sid}` `POST /v0/evidence` `POST /v0/orgs/{orgtag}/sources`
`litmus-ai serve`	`GET /v0/jobs/{job_id}` `POST /v0/surfaces` `POST /v0/worlds`
`litmus-ai compare`	`POST /v0/evidence`
`litmus-ai brief`	none found
`litmus-ai mock`	`POST /v0/surfaces` 2 call(s) that need the workspace’s service token, not an org token
`litmus-ai shell`	2 call(s) that need the workspace’s service token, not an org token
`litmus-ai status`	`GET /v0/bundles` `GET /v0/jobs` 1 call(s) that need the workspace’s service token, not an org token 1 call(s) whose path is built at run time
`litmus-ai export`	none found
`litmus-ai scorecard`	none found
`litmus-ai harbor`	none found
`litmus-ai anon-key`	none found
`litmus-ai pull`	`GET /v0/evidence` `GET /v0/orgs/{orgtag}/sources` `GET /v0/orgs/{orgtag}/sources/{sid}` `GET /v0/whoami` `PATCH /v0/orgs/{orgtag}/sources/{sid}` `POST /v0/evidence` `POST /v0/orgs/{orgtag}/sources/{sid}/pulled`
`litmus-ai labels`	`POST /v0/orgs/{orgtag}/sources/{sid}/labels`
`litmus-ai sources`	`PATCH /v0/orgs/{orgtag}/sources/{sid}`
`litmus-ai org`	`GET /v0/orgs/{orgtag}/archive` 1 call(s) that need the workspace’s service token, not an org token
`litmus-ai target`	none found
`litmus-ai failure-modes`	`POST /v0/orgs/{orgtag}/failure-modes`
`litmus-ai feedback`	`DELETE /v0/orgs/{orgtag}/feedback/{fid}`
`litmus-ai experiment`	`GET /v0/jobs/{job_id}` `GET /v0/orgs/{orgtag}/agents` `GET /v0/orgs/{orgtag}/runs` `GET /v0/runs/{rid}` `GET /v0/whoami` `POST /v0/exports/datadog` `POST /v0/exports/datadog-judges` `POST /v0/orgs/{orgtag}/runs/start`
`litmus-ai mirror`	`GET /v0/orgs/{orgtag}/mirror-in` `GET /v0/orgs/{orgtag}/mirror` `GET /v0/whoami` `POST /v0/orgs/{orgtag}/mirror-in/{kind}`
`litmus-ai mcp`	none found
`litmus-ai eval`	`POST /v0/orgs/{orgtag}/runs/start`

`litmus-ai init`

usage: litmus init --token <token> [--url <workspace>] [--org <org>]

`litmus-ai probe`

usage: litmus probe --mcp "<server command | https://…/mcp>" [--label text] [--read-only] [--header "K: V"] [--seed '{"session_id":"…"}']

`litmus-ai call`

usage: litmus call --mcp "<server|url>" <tool> [--args '{"k":"v"}'] [--session <file>] [--label text]

`litmus-ai record`

usage: litmus record [--wrap bin1,bin2] [--label text] -- <command...>

`litmus-ai push`

usage: litmus push <file|run-dir> [--org <org>] [--label <text>]

`litmus-ai serve`

usage: litmus serve <surface-name> [session-or-trace files...]

`litmus-ai compare`

usage: litmus compare --real "<server|url>" --mock <mock mcp url> [--session <file>] [--header "K: V"] [--push]

`litmus-ai brief`

usage: litmus brief <{'|'.join(BRIEFS)}>

`litmus-ai mock`

usage: litmus mock <name> <traces.json|session.jsonl> [more files...]

`litmus-ai shell`

usage: litmus shell --bundle <name-or-dir> [--world db | --mint] [--knobs JSON]

`litmus-ai export`

usage: litmus export --org <org> --surface <surface> [--out <dir>] [--repo <checkout>]

`litmus-ai scorecard`

usage: litmus scorecard <results-dir | run-dir> [--manifest manifest.json] [--out scorecard.md]

`litmus-ai harbor`

usage: litmus harbor --suite suite --run-dir results/<stamp>-harbor [--task <id>]... [--trials N]

`litmus-ai anon-key`

usage: litmus anon-key show --org <org>      print this org's pseudonym key

`litmus-ai pull`

usage: litmus-ai pull --from datadog-llmobs [--org <org>] [--surface <s>]

`litmus-ai labels`

usage: litmus-ai labels --org <org> --source <src_id> --from langfuse

`litmus-ai sources`

usage: litmus-ai sources set-scope --org <org> --source <src_id> ml_app=<app> [select=<query>]

`litmus-ai org`

usage: litmus-ai org export <slug> [--out <file>]   (else stdout)

`litmus-ai target`

usage: litmus-ai target add <name> --url <workspace> [--token <token> | --token-env <VAR>] [--org <org>] [--use]

`litmus-ai failure-modes`

usage: litmus-ai failure-modes import <file> --org <org>

`litmus-ai feedback`

usage: litmus-ai feedback retire <org> <note id> [<note id> …] [--by <who>]

`litmus-ai mirror`

usage: litmus-ai mirror --org <slug> [--to <url>] [--kinds a,b] [--replace] [--only <id>[,<id>…] | --exclude <id>[,<id>…]] [--dry-run]

`litmus-ai mcp`

usage: litmus-ai mcp [--stdio]

Get started

API reference

Tools

`litmus-ai init`

`litmus-ai probe`

`litmus-ai call`

`litmus-ai record`

`litmus-ai push`

`litmus-ai serve`

`litmus-ai compare`

`litmus-ai brief`

`litmus-ai mock`

`litmus-ai shell`

`litmus-ai export`

`litmus-ai scorecard`

`litmus-ai harbor`

`litmus-ai anon-key`

`litmus-ai pull`

`litmus-ai labels`

`litmus-ai sources`

`litmus-ai org`

`litmus-ai target`

`litmus-ai failure-modes`

`litmus-ai feedback`

`litmus-ai mirror`

`litmus-ai mcp`

​litmus-ai init

​litmus-ai probe

​litmus-ai call

​litmus-ai record

​litmus-ai push

​litmus-ai serve

​litmus-ai compare

​litmus-ai brief

​litmus-ai mock

​litmus-ai shell

​litmus-ai export

​litmus-ai scorecard

​litmus-ai harbor

​litmus-ai anon-key

​litmus-ai pull

​litmus-ai labels

​litmus-ai sources

​litmus-ai org

​litmus-ai target

​litmus-ai failure-modes

​litmus-ai feedback

​litmus-ai mirror

​litmus-ai mcp

`litmus-ai init`

`litmus-ai probe`

`litmus-ai call`

`litmus-ai record`

`litmus-ai push`

`litmus-ai serve`

`litmus-ai compare`

`litmus-ai brief`

`litmus-ai mock`

`litmus-ai shell`

`litmus-ai export`

`litmus-ai scorecard`

`litmus-ai harbor`

`litmus-ai anon-key`

`litmus-ai pull`

`litmus-ai labels`

`litmus-ai sources`

`litmus-ai org`

`litmus-ai target`

`litmus-ai failure-modes`

`litmus-ai feedback`

`litmus-ai mirror`

`litmus-ai mcp`