Overview#
Rolling out a new AI-powered case summarisation feature to three thousand users at once is a risk no engineering team should take. A single edge-case data pattern can cause failures across the entire user base, and reverting requires a deployment cycle. Feature flag management removes that risk: enable the feature for twenty internal testers, expand to a beta cohort, ramp up gradually by percentage, and only proceed to full release when stability and adoption metrics confirm it is ready. If something goes wrong at any stage, a single flag toggle disables it immediately without touching production code.
The Argus Feature Flags Management system provides granular control over platform features at the tenant, user group, and individual user levels. It supports progressive rollouts, A/B testing, beta programme management, and tenant-specific feature configurations across a multi-tenant deployment.
Key Features#
Flag Types and Targeting#
- Boolean toggles for simple on/off feature control
- Percentage-based rollouts for gradual feature exposure across user populations
- User targeting for enabling features for specific individuals or test accounts
- Tenant scoping for organisation-level feature management
- Time-based activation for scheduled feature launches and temporary feature windows
- Geographic and locale-based targeting for region-specific feature availability
Evaluation Hierarchy#
The system resolves flags through a predictable five-level hierarchy: user override takes highest priority, followed by group override, then tenant configuration, then percentage rollout, and finally platform default. This hierarchy ensures consistent, auditable flag resolution across all platform components.
Management and Governance#
- Flag management dashboard with audit logging and change tracking for all modifications
- Dependency management and impact analysis for flag configurations
- Flag lifecycle management from creation through deprecation and removal
- Approval workflows for production flag changes in regulated environments
- Bulk operations for managing related flags across feature sets
Analytics and Monitoring#
- Usage tracking showing feature adoption rates across tenants and user segments
- Adoption metrics measuring how quickly users engage with newly enabled features
- Error correlation identifying whether feature flag changes impact system stability
- Performance impact analysis comparing system behaviour with features enabled and disabled
- A/B test results with statistical analysis for data-driven feature decisions
Feature Organisation#
Feature flags are organised across investigation, analytics, AI, integration, compliance, mobile, and experimental domains. Tagging, categorisation, ownership records, and stale flag detection keep the flag inventory clean and manageable.
Use Cases#
Controlled Rollout. Release features gradually: internal testing, beta users, percentage ramp, full release. Monitor at each stage. Proceed only when metrics confirm readiness.
Tenant Customisation. Enable premium features for licensed organisations, disable irrelevant modules, configure integrations specific to each tenant's operational environment, and manage trial features for evaluation periods.
Beta Programme Management. Enrol beta users and organisations, enable experimental features, collect structured feedback, track usage metrics, and graduate features to production based on quality and adoption thresholds.
Incident Response. Disable a problematic feature in production without a code deployment when issues are detected, limiting user impact while engineering investigates. Target rollback to specific tenants or user groups.
Integration#
- Connects with tenant management and configuration systems for organisation-level feature control
- Integrates with access control and permission systems for role-based feature gating
- Links to platform configuration and system settings for centralised administration
- Works with release management and deployment workflows for coordinated feature launches
- Compatible with monitoring and alerting systems for feature impact tracking
- Supports analytics platforms for feature adoption and A/B test analysis
- Feeds into compliance reporting for feature audit trails and change documentation
Open Standards#
- GraphQL (June 2018 specification): All feature flag queries and mutations are served via a typed GraphQL API, enabling structured, self-documenting flag evaluation requests from any compliant client.
- JSON Web Token / RFC 7519: Every flag evaluation request must carry a valid RS256 JWT; the service verifies the token signature and extracts role and tenant claims to determine the correct evaluation path.
- JSON Web Key Sets / RFC 7517: The flag evaluation gate resolves RS256 public keys dynamically from a JWKS endpoint, allowing key rotation without redeployment.
- OAuth 2.0 / RFC 6749: Permission scopes embedded in OAuth 2.0 access tokens drive the ontology access check that guards the feature flag read endpoint, ensuring only authorised callers receive flag state.
- JSON / RFC 8259: Feature flag state is stored, cached, and transmitted as plain JSON objects mapping domain names to boolean values, making the payload trivially consumable by any standards-compliant HTTP client.
- Role-Based Access Control (NIST SP 800-207 / ANSI INCITS 359): The five-level evaluation hierarchy enforces RBAC by resolving flags differently for superusers, SI admins, tenant admins, and standard users, with each role tier constraining the set of features the next tier may enable.
Last Reviewed: 2026-02-05 Last Updated: 2026-04-14