Glossary
Canonical terms used throughout the Agent Stream documentation.
Agent Stream is the model-governance layer for AI agents: it sits between callers and the LLM providers, APIs, and tools they call, and applies routing, guardrails, budgets, and observability to every execution from one configuration layer. It is part of the Affinidi Trust Fabric.
This glossary provides clear, consistent terminology to help you get started quickly.
Routing model
| Term | Meaning |
|---|---|
| Surface | The single configuration and runtime unit Agent Stream uses to represent one governed connection. Surfaces → |
| LLM Surface | A surface configured against one LLM provider. Carries the full guardrail, routing, resilience, and cost pipeline. |
| IDE Surface | A governed, per-caller-filtered model menu that aggregates one or more LLM Surfaces behind a single OpenAI-compatible endpoint, for editors and OpenAI-compatible agents. IDE Surface → |
| Variant | A named overlay on a surface, addressable at {route}$alias/..., that overrides part or all of the surface’s behaviour (provider, model, guardrails, limits) without a separate surface record. Variants and progressive rollout → |
| Access Point node | The canvas node (labelled Access Point) on any surface, LLM or IDE, that defines how incoming requests reach it: listen address, path prefix, and custom path, composed into the surface’s route. Distinct from a connection point, which is the inbound DIDComm endpoint used for gateway-to-gateway traffic. |
| LLM node | The canvas node (labelled LLM) on an LLM Surface that defines the provider, model, and API key secret the surface calls out to. |
| Mode | The LLM node’s operating mode: LLM Direct (the default, one model), Decide LLM based on content (a Decider picks one candidate), Compare output from multiple LLMs (a Comparer returns every candidate’s answer), or Summarise output from multiple LLMs (a Summariser merges them). Pipeline and stages → |
| Router | A canvas element whose ordered routing rules send a request to a variant by header, body field, or attachment type. An explicit $alias in the route takes precedence over a rule, and a rule over a canary split. Inputs and modalities → |
| Route | The composed full URL shown on a surface’s Access Point node: listen address, path prefix, and custom path combined. Callers send requests to this address. |
Providers and wire formats
| Term | Meaning |
|---|---|
| Provider | The LLM backend an LLM Surface reaches: OpenAI, Anthropic, Google Gemini, AWS Bedrock, Microsoft 365 Copilot, Ollama, Groq, xAI (Grok), OpenRouter, Fireworks AI, Cerebras, DeepSeek, Moonshot AI (Kimi), SambaNova Cloud, Hyperbolic, NVIDIA NIM, or Nebius AI Studio. Azure OpenAI is listed in the provider catalogue but isn’t available on this appliance. |
| Model catalogue | The hot-reloadable source of provider pricing, capabilities, and request profiles. A new model becomes available through the dashboard, not a code change. |
| Client API format | The request and response format (wire format) a client uses to call a surface: OpenAI Chat Completions, Anthropic Messages, or OpenAI Responses. Every LLM Surface accepts all three by default, set in its Client API formats section. Independent of which provider the surface itself is configured to reach. |
| Inputs | The palette group of elements that opt a surface in to attachments: Audio File, Image File, Document File, and Audio Stream for realtime voice. Audio and images are rejected on a surface without the matching element. Inputs and modalities → |
| Starter template | A pre-built canvas layout offered when creating a surface, such as OpenAI Chat Completions, which can pre-wire a provider, model, and endpoint. |
| Partial template | A template that merges a bundle of elements onto an existing surface instead of replacing it, such as the Guardrails Pack. |
Guardrails
| Term | Meaning |
|---|---|
| Guardrails | The layered content-safety pipeline applied to a surface’s traffic. Guardrails → |
| Prompt Guard | Deterministic pattern matching plus model-backed AI detection for PII, run on the request (both) and response (pattern only) seams. Each match can be rejected, redacted, or replaced with a NER ID. |
| NER ID | A stable pseudonym assigned to a detected PII value, preserving relationships between values without exposing the raw data. PII protection and NER IDs → |
| Expert Witness | A purpose-trained external safety classifier (for example EnkryptAI, Lakera Guard, Azure Prompt Shields, Azure Content Safety) consulted on the request seam, the response seam, or both. |
| Judge | An LLM-based or moderation-based pre-check on the request that can block it before the main model is called. In Decide LLM based on content mode on a buffered request, it runs in parallel with the Decider. Otherwise it completes first. |
| Jury | A multi-juror LLM post-review of the model’s response that can approve, block, or regenerate an answer, with optional per-juror retry. |
| Fail closed | The default for the two guardrails that offer a fail-open or fail-closed choice, Expert Witnesses and the AI Matching detector: if the check can’t run, the request is blocked rather than allowed through unchecked. Other guardrails have no fail-open setting. |
| Rejection taxonomy | The stable set of categories (hate, toxicity, prompt injection, jailbreak, PII, secrets, policy violation, and more) every guardrail block is classified into, with Uncategorized as the fallback, charted per team, member, surface, and stage. |
Access control
| Term | Meaning |
|---|---|
| OPA (Open Policy Agent) | The embedded Rego policy engine, evaluated in-process with no external OPA server required. OPA policies → |
| Gateway-level policy | The Trust Fabric policy type, attached to the appliance itself. It gates fabric (gateway-to-gateway, DIDComm) traffic rather than LLM Surface requests. |
| Surface-level policy | Policy attached to one surface. On an LLM Surface request, it is evaluated after the appliance-wide (global) policies for that surface type, and a deny from either blocks the request, unless the global policy is set to Monitor only. |
| RBAC role ladder | The three roles governing dashboard access: Administrator (the top of the ladder), Power User (adds mediators and connection points on top of the User role, still with no secrets or user/RBAC management), and User (the default role, with broad view access plus edit and delete on LLM and IDE Surfaces and notifications; no secrets, user/RBAC, or settings access). Security and access control → |
| T&C Manager | The Settings tab, for administrators, that shows the Affinidi Terms and publishes your own terms. Users accept the current terms with I agree to … checkboxes at registration, and on the Review the Terms screen at sign-in when a new version needs acceptance. Settings → |
| Secrets store | The encrypted store holding provider API keys and other credentials, referenced by ID from a surface rather than embedded in configuration. |
| Source authentication | Verification of the inbound caller’s identity, using one of the JWT Bearer, API Key (secret store), API Key (provider), or mTLS (Mutual TLS) methods. |
Cost, attribution, and monitoring
| Term | Meaning |
|---|---|
| Usage limit | A hard ceiling, per surface, per stage, or per juror, on monthly USD spend or monthly token usage. Once spend or tokens reach it, further requests are blocked with 429 before they reach the provider. The request that crosses it is itself allowed through. |
| Breach alert | A notification, not a block: fires after a cost or token threshold is crossed, or an anomaly is detected, so the request still completes. |
| Member | A registered caller, a person or a client, that reaches models through the appliance. |
| Team | A group of members. Usage is attributed to both the member and its team on any surface where attribution is enabled. |
| Virtual key | The effective behaviour of a member: a model allow-list, per-minute rate limits, and monthly ceilings, on top of a surface’s own limits. Teams and attribution → |
| LLM Drift | The Monitoring page that compares a surface’s behaviour dimensions, such as refusal rate, latency, and cost, with a pinned baseline and marks each stable, warn, or drift. LLM drift detection → |
| Drift Harness | The enhancement on a surface’s LLM node that switches on drift monitoring and samples live traffic for monitoring and replay. |
| Baseline | The snapshot of a surface’s drift dimensions pinned with Set baseline, which later windows are compared against. |
| Replay evaluation | A run that re-executes captured sessions through the same or another surface or variant and compares each new answer with the original. |
| x-agent-stream- headers* | Per-request response headers carrying trace ID, provider, model, token counts, cached-token count, cost, cache hit or miss, and ingress wire format. |
| Governance record | A signed W3C Verifiable Credential produced for each interaction the surface’s Record setting covers, when recording is enabled, capturing the prompt, every stage’s verdict, the completion, and metrics, chained to the previous record on the same surface. Governance records → |
Resilience
| Term | Meaning |
|---|---|
| Failover | Routing to ordered backup providers when the primary is unreachable or erroring. |
| Load balancing | Distributing traffic across a pool of providers by weighted round-robin, lowest observed latency, or lowest price. A surface uses either failover or load balancing. |
| Response cache | A cache serving a repeat or near-duplicate prompt without calling the provider, in exact or semantic (embedding-similarity) mode. A streamed request can also be served from the cache, and a streamed answer is written to the cache only when the request offers no tools. |
Trust fabric
| Term | Meaning |
|---|---|
| G2G (gateway-to-gateway) | A call from this appliance to a connected peer appliance over DIDComm v2.1, used to browse and buy models from the peer’s model marketplace. Trust-fabric integration → |
| Connection point | The DIDComm endpoint another appliance uses to establish and use a connection with this one. |
| Mediator | A component that routes DIDComm messages between instances that cannot connect directly. |
| x402 | An HTTP 402 payments protocol for blockchain-settled, metered access to a model or service. |
| Model marketplace | A built-in catalogue and purchase flow letting a caller provision an API key for a purchased model directly from the appliance. |
Was this page helpful?
Glad to hear it! Please tell us how we can improve more.
Sorry to hear that. Please tell us how we can improve.
Thank you for sharing your feedback so we can improve your experience.