# CIVEDRA CLI

Evaluate any AI model's trust and safety status
in under 2 seconds from your terminal.

```bash
civedra check deepseek-r1
civedra check claude-3-5-sonnet
civedra check gpt-4o

Installation

npm install -g @civedra/cli

Requires Node.js 14 or higher.

Configuration

The CLI works out of the box with no configuration.

To use a custom endpoint (for enterprise deployments):

export CIVEDRA_HOST=your-endpoint.com export CIVEDRA_KEY=your-key civedra check deepseek-r1

Usage

# Check any AI model
civedra check <model-name>

# Examples
civedra check deepseek-r1
civedra check claude-3-5-sonnet
civedra check llama-3-70b
civedra check gpt-4o

Example Output

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
CIVEDRA TRUST CHECK
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Model:     DeepSeek-R1
Developer: DeepSeek (China)
CTI Score: 12/100
Verdict:   ⛔ PROHIBITED

!! DEPLOYMENT PROHIBITED !!
NDAA FY2026 §1532 — Covered Nation: China
This model cannot be deployed in any
DoD or IC environment under current law.

Full evaluation: civedra.com/model/deepseek-r1
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Exit Codes

For CI/CD pipeline integration:

| Code | Meaning | |------|---------| | 0 | AUTHORIZED — safe to proceed | | 1 | PROHIBITED or CONDITIONAL — block deployment | | 2 | Model not found in registry | | 3 | API error |

GitHub Actions Integration

- name: CIVEDRA AI Model Check
  run: |
    npm install -g @civedra/cli
    civedra check ${{ env.AI_MODEL }}

About

CIVEDRA is an AI Integrity and Trust Intelligence Platform built for national security environments. ARBITER continuously evaluates AI models for provenance, covered-nation origin, behavioral integrity, and NDAA §1532 compliance.

civedra.com | Trust Index | API Docs

License

MIT — see LICENSE file.