CLI Reference

The mima CLI is installed with the Python SDK:

$
pip install mima-governance
$
mima --version

Or run via Python:

$
python -m mima_governance --version

Authentication

mima login

Store your API key and workspace credentials locally.

$
mima login

Prompts for your API key, then resolves and stores your workspace ID. Credentials are written to ~/.mima/config.json with owner-only permissions (0600).

$
mima login --api-key mima_ext_...

Evidence

mima push

Push a GRC evidence record from the command line. Pass the record type as the first positional argument, then supply fields as named flags.

$
mima push change_event \
>
  --by ci-bot@company.com \
>
  --description "Deploy v1.2.3 to production" \
>
  --environment production \
>
  --system api-service

$
mima push ai_risk_assessment \
>
  --system loan-scorer \
>
  --risk-tier limited \
>
  --use-case "Customer support routing" \
>
  --intended-purpose "Route queries to the correct team" \
>
  --impact-domains customer_service \
>
  --art5-self-assessment true \
>
  --assessor alice@example.com

$
mima push model_evaluation \
>
  --model-id loan-scorer-v2 \
>
  --dataset holdout-q2-2026 \
>
  --accuracy 0.94 \
>
  --evaluated-by alice@example.com \
>
  --evaluation-type quarterly

Pipe JSON for scripted use:

$
echo '{"by":"ci-bot@company.com","description":"Deploy","environment":"production","system":"api-service"}' \
>
  | mima push change_event --stdin

Add --dry-run to preview controls without writing:

$
mima push model_evaluation \
>
  --model-id loan-scorer-v2 --dataset holdout --accuracy 0.94 \
>
  --evaluated-by alice@example.com --dry-run
$
# Output: Would earn: EUAIA_ART9_4, ISO42001_9_1

Common flags

Environment variable: MIMA_BASE_URL

Used by mima push and mima gates. Set when pointing at a self-hosted deployment:

$
export MIMA_BASE_URL=https://mima.internal.example.com/api
$
mima push ...

Gates

mima gates

Check whether required evidence gates pass for a system.

$
mima gates --system my-pipeline

Outputs a table of gate name, pass/fail status, and blocking status. Exits with code 0 if all gates pass, 1 if any soft gates fail, 2 if any hard gates fail.

Use in CI to block deploys when gates fail:

$
mima gates --system my-pipeline || exit 1

Options

Controls

mima derive-controls

Get recommended evidence record types for a system description.

$
mima derive-controls --description "A classification model that routes \
>
  customer support tickets to the correct team."

Returns a list of recommended record types and the controls they earn.

Recommended records for this system:
  ai_risk_assessment     → EUAIA_ART9, ISO42001_6_1
  model_evaluation       → EUAIA_ART9_4, ISO42001_9_1
  human_oversight        → EUAIA_ART14, ISO42001_6_1_2
  training_data_governance → EUAIA_ART10

Options

Environment variable: MIMA_API_URL

Used by mima derive-controls. Defaults to https://api.mima.ai/api.

Status

mima status

Show your current posture score and top open gaps.

$
mima status

Posture score: 74/100
Open gaps (top 5):
  [HIGH]   model_drift_event not logged in 90 days (loan-scorer)
  [MEDIUM] training_data_governance missing for 2 systems
  [MEDIUM] quarterly model_evaluation overdue (customer-support-ai)
  [LOW]    vendor_risk not reviewed in 180 days (OpenAI)
  [LOW]    governance_review overdue
Gates: 3/5 passing

Configuration file

Stored at ~/.mima/config.json. Do not edit manually — use mima login.

1
{
2
  "api_key": "mima_ext_...",
3
  "workspace_id": "ws_...",
4
  "base_url": "https://api.mima.ai"
5
}

Permissions are set to 0600 (owner read/write only). The API key is never written to .env files or environment exports by mima login.

Exit codes