CLI Command Reference

Global flags

Flag	Description
--api-key	GOVERN API key (overrides env/credentials)
--org-id	Organization ID
--output	Output format: table (default), json, yaml
--quiet	Suppress non-essential output
--debug	Enable debug logging
--config	Path to CLI config file (default: ~/.govern/config.yaml)

---

govern assess

Assess one or more inference inputs against GOVERN policy.

govern assess [flags]

Flags

Flag	Required	Description
--model	Yes	Model ID (e.g., claude-sonnet-4-20250514)
--prompt	Yes*	Inline prompt string
--prompt-file	Yes*	Path to JSON prompt file
--response	Yes*	Inline response string
--response-file	Yes*	Path to response text file
--batch-file		JSON file containing array of prompt/response pairs
--mode		Override mode: log, flag, block
--output		table, json, yaml
--fail-on-flag		Exit code 1 if any assessment is flagged
--fail-on-block		Exit code 1 if any assessment would be blocked

*Either --prompt/--response (inline) or --prompt-file/--response-file (file) required.

Examples

# Assess a single inference
govern assess \
  --model claude-sonnet-4-20250514 \
  --prompt "Summarize the Q3 report" \
  --response "Q3 revenue was $4.2M, up 18% from Q2..."

# Output:
# Assessment: PASS
# Security:  0.01  ✓
# Bias:      0.00  ✓
# Accuracy:  0.87  ✓
# Drift:     0.03  ✓
# Cost:      0.05  ✓

# Assess from files
govern assess \
  --model claude-sonnet-4-20250514 \
  --prompt-file tests/prompt.json \
  --response-file tests/response.txt \
  --output json

# Batch assessment (CI mode)
govern assess \
  --batch-file tests/govern/test-cases.json \
  --fail-on-flag \
  --output json

Batch file format

[
  {
    "id": "test-001",
    "model": "claude-sonnet-4-20250514",
    "prompt": [{"role": "user", "content": "What is 2+2?"}],
    "response": "2+2 equals 4.",
    "metadata": {"test_name": "basic-math"}
  }
]

---

govern policy

Manage your organization’s governance policy.

govern policy <subcommand>

govern policy show

govern policy show
# Shows current policy: mode, thresholds, enabled scorers

govern policy set-mode

govern policy set-mode block
# Sets org-wide scoring mode to 'block'

govern policy set-threshold

govern policy set-threshold --scorer security --threshold 0.65

---

govern events

Query governance events from the audit trail.

govern events list [flags]

Flag	Description
--since	Start time (e.g., 24h, 7d, 2026-04-01)
--action	Filter by action: pass, flag, block
--scorer	Filter by scorer that triggered
--model	Filter by model
--limit	Max events to return (default: 50)

# Last 24 hours of flags
govern events list --since 24h --action flag

# Exports as JSON
govern events list --since 7d --output json > events.json

---

govern probes

Manage and inspect connected Probe instances.

govern probes list
govern probes status <probe-id>
govern probes reset-drift <probe-id>
govern probes push-policy <probe-id>

---

govern auth

Authentication management.

govern auth login        # Interactive browser login
govern auth logout       # Remove stored credentials
govern auth status       # Show current auth state
govern auth set-key KEY  # Set API key directly

---

govern version

govern version
# govern version 1.2.0 (linux/amd64, commit abc123, built 2026-04-01)

Exit codes

Code	Meaning
0	Success
1	Assessment failure (with --fail-on-flag or --fail-on-block)
2	Authentication error
3	Invalid input
4	Network error
5	Policy error