Skip to main content
Forge

Security

The control model behind every workflow

This page describes the security capabilities a platform like Forge is designed around. It is written as template demo content — it documents product capabilities, not audited certifications.

Template demo content. The capabilities below describe how a platform of this kind is designed. Forge is a fictional company and holds no certifications, attestations or audits of any kind. Replace this page with your own verified security documentation before publishing.

Encryption

Data protected in transit and at rest, with key handling that does not depend on a workflow author getting it right.

  • TLS 1.3 for all external and inter-service traffic
  • AES-256 encryption for data at rest, including run payloads
  • Envelope encryption with regular key rotation
  • Secrets held in a managed vault and referenced, never inlined

Access control

Every action is attributable to an identity, and every identity has the narrowest scope that still lets it work.

  • Role-based access scoped to workspace, project or single workflow
  • Named approvers required before a workflow revision is published
  • Scoped API credentials with independent rotation and revocation
  • Break-glass access recorded as a distinct, alertable audit event

SSO & provisioning

Identity comes from your directory, so joiners and leavers are handled where they are already handled.

  • SAML 2.0 and OIDC single sign-on
  • SCIM provisioning and de-provisioning
  • Directory groups mapped to platform roles
  • Enforced session lifetimes and re-authentication for sensitive actions

Audit logging

A complete, immutable record of what happened — designed to be read by someone who was not there.

  • Immutable events for publish, run, override and configuration change
  • Actor, timestamp, workflow revision and affected records on every event
  • Retention windows configurable per workspace
  • Streaming export into your own log estate or GRC tooling

Data handling

The platform holds as little as it can, for as short a time as you allow.

  • Field-level masking applied before data reaches an intelligence step
  • Configurable processing and storage residency per workspace
  • Retention windows with verified deletion at expiry
  • Read-through connectors so records stay in your systems of record

Platform resilience

Durable execution means a failure degrades throughput rather than losing work.

  • Checkpointed runs that resume after an instance restart
  • Multi-zone deployment within each configured region
  • Tested backup and restore procedures for control-plane state
  • Dead-letter queues with replay for anything that exhausts retries

Incident response

What happens when something goes wrong

An incident process is only credible if it is written down before it is needed. This is the shape of the process a platform of this kind should operate.

  1. 01

    Detect

    Automated alerting on availability, error rate and anomalous access patterns, backed by a 24/7 rotation.

  2. 02

    Contain

    Isolate affected components, revoke implicated credentials and stop the spread before diagnosing the cause.

  3. 03

    Notify

    Communicate to affected customers within the contractual window, with facts rather than reassurance.

  4. 04

    Resolve

    Restore service, verify integrity of affected runs and replay anything that needs replaying.

  5. 05

    Review

    Blameless post-incident review with published corrective actions and owners.

Practices

Engineering practices

The engineering habits that keep the control model true in day-to-day work.

  • Peer review required on every change to production systems
  • Dependency and container scanning in continuous integration
  • Independent penetration testing on an annual cadence
  • A published vulnerability disclosure policy with a response commitment
  • Least-privilege infrastructure access with time-bound elevation
  • Security review required before a new connector reaches general availability

Security contact

Reporting a vulnerability

In a production deployment of this template, security reports should go to a monitored address with a published response commitment and a disclosure policy. The address below is a demo placeholder and is not monitored.

Security contact

security@forge.example.com

Response

Illustrative commitment: acknowledgement within one business day.

Contact us

Security review

Bring us your questionnaire

We would rather answer the hard questions early than discover them in week six of a procurement cycle.

Talk to usSee enterprise plans