@interven/policy-cli

Policy-as-Code CLI for Interven. Validate, diff, and apply YAML/JSON policy files against your tenant. Designed for Git-managed policy workflows + CI/CD.

Install

npm install -g @interven/policy-cli

Or run on demand:

npx @interven/policy-cli validate policies/

Usage

interven-policy validate policies/*.yaml         # syntax + schema check
interven-policy diff     policies/*.yaml         # show drift vs server
interven-policy apply    policies/*.yaml         # idempotent upsert
interven-policy fmt      policies/*.yaml         # canonical formatting

Policy YAML shape

The CLI accepts two interchangeable shapes — pick whichever you prefer. validate, diff, and apply all work identically with either; the simple shape is converted to the canonical shape internally at load time.

Shape 1 — Simple (recommended)

One file can hold many policies under a top-level policies: list. This is the shape used by all the starter packs:

policies:
  - name: secrets-egress-block
    description: Block any outbound Slack/email/webhook carrying SECRETS.
    rules:
      - rule_id: slack-secrets-deny
        action: DENY            # ALLOW | DENY | SANITIZE | REQUIRE_APPROVAL
        priority: 100
        match:
          tool_name: slack
          operations: ["post_message", "upload_file"]
        condition:
          has_data_class: ["SECRETS"]

Shape 2 — Canonical / K8s-style (engine-native)

apiVersion: interven.security/v1
kind: Policy
metadata:
  name: secrets-egress-block
  description: Block any outbound Slack/email/webhook carrying SECRETS.
  env: production           # optional; defaults to CLI's --env
spec:
  rules:
    - rule_id: slack-secrets-deny
      description: Slack post containing SECRETS → DENY
      action: DENY           # ALLOW | DENY | SANITIZE | REQUIRE_APPROVAL
      priority: 100
      match:
        tool_name: slack
        operations: ["post_message", "upload_file"]
      condition:
        has_data_class: ["SECRETS"]

Field reference:

| Section | Field | Notes | |---------|-------|-------| | match | tool_name | Exact match against tool (slack, github, ...) | | match | operations | Patterns: exact name, *, method:GET, category:foo | | match | verbs | READ / WRITE / ADMIN | | match | org_patterns / repo_patterns / channel_patterns | Wildcard supported | | condition | has_data_class | Array of SECRETS / PII / INTERNAL | | condition | is_external_principal | Boolean | | condition | volume_burst | Boolean (agent >20 ops/hr) | | condition | body_contains_any / body_contains_all | Substring patterns in body |

Legacy flat shape also accepted:

# Older docs used this — gateway transparently normalizes:
- decision: DENY
  data_classes: ["SECRETS"]
  tool_name: slack

The CLI's validate accepts both. apply sends them as-is; the gateway translates flat → nested before storing.

See templates/policies/ for production-grade examples (HIPAA, SOC2, PCI-DSS, GDPR).

CI/CD

# .github/workflows/policies.yml
on: [push, pull_request]
jobs:
  policy-ci:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - run: npx @interven/policy-cli validate policies/
      - if: github.event_name == 'pull_request'
        run: npx @interven/policy-cli diff policies/ --env staging
        env:
          INTERVEN_API_KEY: ${{ secrets.INTERVEN_API_KEY_STAGING }}
      - if: github.ref == 'refs/heads/main'
        run: npx @interven/policy-cli apply policies/ --env production
        env:
          INTERVEN_API_KEY: ${{ secrets.INTERVEN_API_KEY_PROD }}

Env vars

| Var | Default | Purpose | |-----|---------|---------| | INTERVEN_API_KEY | required for diff/apply | Bearer token (parent iv_live_*) | | INTERVEN_GATEWAY | https://api.intervensecurity.com | Override for self-hosted |

License

MIT