Every Cognis Neural Suite tool is built to drop into an existing enterprise stack. This guide lists the supported integration surfaces. Where a surface is marked planned, the interface exists and contributions are welcome (see CONTRIBUTING.md).
| Surface | How | Status |
|---|---|---|
| CLI / exit codes | --fail-on <severity> for CI gates; JSON on stdout |
✅ |
| JSON / SARIF | machine-readable findings; SARIF for code-scanning | ✅ |
| MCP server | <tool> mcp exposes capabilities to agents/Cognis.Studio |
✅ |
| REST / Webhooks | integrations/webhook.py posts findings to any endpoint |
✅ |
| Identity — SSO | SAML 2.0 / OIDC (Okta, Entra ID, Auth0, Google, Ping) | planned |
| Identity — SCIM | user/group provisioning | planned |
| SIEM | Splunk HEC, Microsoft Sentinel, Elastic, Chronicle, Datadog | planned |
| Ticketing | Jira, ServiceNow, Linear, GitHub/GitLab Issues | planned |
| ChatOps | Slack, Microsoft Teams, Discord, PagerDuty | planned |
| Cloud | AWS, Azure, GCP (read-only roles, EventBridge/PubSub) | planned |
| Data | S3/GCS/Azure Blob, Postgres, Snowflake, BigQuery, Kafka | planned |
| Secrets | Vault, AWS/Azure/GCP secret managers, 1Password | planned |
| GRC | export to the Cognis compliance tools (soc2box, frameworkmap) | ✅ |
# CI gate (GitHub Actions, GitLab CI, Jenkins, etc.)
<tool> scan . --format sarif --out results.sarif --fail-on high
# Stream findings to a webhook / SIEM forwarder
<tool> scan . --format json | python integrations/webhook.py --url "$COGNIS_WEBHOOK_URL"
# Use from an AI agent over MCP
<tool> mcpIntegrations read from environment variables (12-factor) or cognis.toml:
[integrations.siem]
type = "splunk" # splunk | sentinel | elastic | datadog
endpoint = "https://hec.example.com:8088"
token = "env:COGNIS_SIEM_TOKEN"
[integrations.ticketing]
type = "jira"
project = "SEC"See integrations/ for adapter stubs you can extend.