Product Feature Inventory

This inventory is the documentation checklist for the product catalog in lib/substrate-product-categories.ts. Each product area below lists the capabilities currently declared in the website product model.

AI

Inference

• llama.cpp and vLLM runtimes
• Model download and integrity verification
• Node enrollment, health, drain, and updates
• Private and public capacity planes
• Model and runtime attestation
• Distributed KV cache sharing
• Prefill/decode and distributed-runtime foundations

Distributed KV caching

• Cross-worker KV cache sharing
• Prefix-aware cache placement
• Cache-affinity routing
• Shared cache lifecycle
• Tenant and model isolation
• Cache hit and reuse evidence

Marketplace

• Model and artifact discovery
• Private capacity pools
• Public surplus-capacity offers
• Skill search, import, and updates
• Signed usage records
• Tenant, project, model, and cache isolation

Agent fleet

• Single and multi-agent execution
• Validated task plans and concurrent waves
• Agent identities, keys, sessions, and connections
• Cohorts and fleet-level operations
• Human approval and secure forms
• Operational memory and verification loops

AI gateway

• OpenAI-compatible Chat and Responses APIs
• Cost, latency, throughput, and balanced routing
• Provider and model fallback
• Exact response cache and cache affinity
• PII, injection, exfiltration, and tool controls
• Durable usage, request history, and traces

RAG

• Schema and entity discovery
• Lexical and vector retrieval
• Relationship and lineage signals
• Permission-aware context
• Sensitive-data redaction
• Agent and AI gateway integration

Semantic layer

• Schema and entity discovery
• Relationship and lineage graphs
• Entity resolution
• Weighted retrieval and embeddings
• Curated context briefs
• Schema-search tools and prompt augmentation

Training

• Dataset registration and lineage
• Training-job orchestration
• LoRA and adapter lifecycle
• Checkpoint and artifact management
• Model evaluation and approval gates
• Deployment into governed model routes

Storage

Key-value

• High-throughput reads and writes
• Low-latency key access
• Workload-aware caching
• Policy-aware gateway paths
• Migration and replay readiness
• Self-hosted deployment

Relational

• SQL and relational access
• Schema-aware operations
• Transactions and composition
• Workload intelligence
• Live migration paths
• Self-hosted deployment

Object storage

• Durable object storage
• Bucket and object policy
• Placement-aware access
• Artifact lifecycle
• Migration and replication paths
• Self-hosted deployment

Document

• Document-native access
• Schema discovery
• Collection and field policy
• Query and workload evidence
• Live migration paths
• Self-hosted deployment

Streaming

• High-throughput event streams
• Topic and consumer controls
• Replayable event history
• Gateway and storage integration
• Operational telemetry
• Self-hosted deployment

Data in Motion

Multi-gateway

• Endpoint registration and metadata
• PostgreSQL, MongoDB, Redis, SQL, NoSQL, cache, vector, warehouse, and API paths
• Connection pooling and backend lanes
• Command, query, and endpoint policy
• Traffic routing, mirroring, and failover
• Scoped credentials and access evidence

Auto caching

• Repeat-read detection
• Policy-aware cache placement
• Freshness and invalidation controls
• Hot-path optimization
• Cache hit and miss evidence
• Per-tenant and per-endpoint isolation

Traffic replication

• Multi-endpoint traffic copies
• Source-authoritative execution
• Response and error comparison
• Target health and latency evidence
• Cohort and route controls
• Migration and cutover preparation

APIs & integrations

• Published API lifecycle
• Endpoint and template bindings
• Function endpoints
• REST and GraphQL integrations
• SaaS and developer-tool endpoints
• Migration-aware execution

Workflows

• Workflow definitions and versions
• Workflow planning and runs
• Workflow lifecycle controls
• Triggers and event sources
• Reusable templates
• Operational status and evidence

Transactions & composition

• Cross-database ACID transactions
• Typed template inputs
• Conditions and iteration
• JSON mapping utilities
• Template access controls
• Versioned execution

Modernization

Live migrations

• Historical copy
• Live write capture
• Log-and-replay
• Progress and throughput metrics
• Pause and resume
• Completion and cancellation controls

Modernization

• PostgreSQL, MongoDB, and MySQL selected routes
• Schema and field mappings
• Request mappings
• Compatibility reports
• Vector and search migration previews
• Route-specific maturity controls

Migration readiness

• Source and target preflight
• Compatibility analysis
• Historical workload analysis
• Readiness recommendations
• Route and mapping plan
• Approval-ready evidence

Traffic replay & live testing

• Traffic mirroring
• Captured request replay
• Controlled live traffic
• Canary and cohort targeting
• Source-target verification
• Performance thresholds and fail-closed controls

Backup & recovery

• Backup lifecycle
• Snapshot lifecycle
• Artifact download
• Status and statistics
• Recovery primitives
• Migration-linked evidence

Operations

Live operations

• Operational overview
• Live event streams
• Endpoint and fleet status
• Migration and agent activity
• Notifications and triggers
• Cross-system evidence

Workload intelligence

• Hot and cold analysis paths
• Query and command fingerprints
• Anomaly state transitions
• Adaptive metadata polling
• Discovery and burst windows
• Pattern, sequence, and aggregate analysis

Recommendations & optimization

• Hot keys and oversized values
• TTL and pipeline opportunities
• Expensive query and fanout detection
• Error, stale, and unusual-pattern findings
• Cost and route optimization
• Semantic annotations and agent triggers

Incidents & investigations

• Anomaly timelines
• Investigation runs
• Agent trigger rules
• Evidence capture
• Approval and verification
• Operational memory

Fleet & runtime health

• Endpoint and node health
• Connection and session state
• Model and cache availability
• Placement and route evidence
• Load balancing and failover
• Drain, update, and recovery state

Costs & usage

• Request and token usage
• Model and provider cost
• Agent and key attribution
• Migration throughput
• Capacity settlements
• Budgets and rate limits

Logs, metrics & traces

• OTLP logs and traces
• Prometheus metrics
• ClickHouse analytics
• Request and route traces
• Streaming AI telemetry
• Core, IAM, endpoint, gateway, migration, and fleet metrics

Customer infrastructure

• Customer infrastructure provisioning
• Data and runtime placement
• Fleet and node enrollment
• Upgrade and drain
• Provider adapters
• Status, entitlement, and recovery workflow

Security

AI & agent security

• Prompt-injection and exfiltration detection
• Tool allowlists and approval
• Model and route policy
• Agent access boundaries
• Token budgets and rate limits
• Run evidence and verification

Data access security

• Server-side credential injection
• Encrypted secret storage
• Endpoint-level security
• Per-user backend credentials
• Command policy
• Data masking and fail-closed execution

Universal masking

• Developer-safe data views
• Agent context masking
• Field and attribute policies
• Role and purpose-aware rules
• Consistent gateway enforcement
• Masking evidence and audit

PII & data loss prevention

• Request and response detection
• Organization PII dictionaries
• Template-derived sensitive terms
• Redaction and blocking
• Endpoint masking
• DLP scan and audit metadata

Identity & access

• Human and agent authentication
• OIDC identity providers
• Control-plane and data-plane RBAC
• Auth groups
• Scoped API and gateway keys
• Session history and revocation

Policy & governance

• Content policy profiles
• Policy bindings
• Access boundaries
• Trusted-principal rules
• Policy simulation
• Governance snapshot hydration

Approvals & audit

• Approval requests and delegation
• Tool and agent evidence
• Audit activity
• Evidence-pack templates
• Lineage and feature records
• Governance exports and replay jobs

Event log

• Per-command audit events
• Governance feature history
• Agent run events
• Decision and approval records
• Trace and resource correlation
• Retention, filtering, and export

VPN & NAT gateway

• VPN networks and peers
• NAT gateway egress
• Private routes
• Configuration rendering
• Endpoint-node placement
• TLS and mTLS
• Fail-closed transport and credential behavior