lAItest / AI field guide · edition 2026-09-26
Know the language. Build with clarity.
Plain-English definitions for the concepts, patterns, protocols, and tools behind modern AI work.
414 defined terms · 16 topic areas · 50 essentials
Security & reliability terms
27 matching terms.
- Durable execution Concept · Security & reliability
Recover and continue a workflow across interruptions using persisted state or event history.
- Idempotency Concept · Security & reliability
Repeated execution of the same operation has the same intended effect as one execution.
- Least privilege Concept · Security & reliability
Grant only the access and capabilities necessary for a specific task.
- Prompt injection Concept · Security & reliability
Untrusted content attempts to redirect a model away from the application's intended instructions.
- Approval gate Concept · Security & reliability
Require authorization before a sensitive action proceeds.
- Audit log Concept · Security & reliability
A durable record of actions and relevant identity, authorization, and outcome information.
- Budget enforcement Concept · Security & reliability
Mechanically limit execution by time, tokens, money, iterations, or allowed actions.
- Cancellation Concept · Security & reliability
A request to stop work; already-completed external side effects may require separate compensation.
- Checkpointing Concept · Security & reliability
Persist execution state so work can resume from a known boundary.
- Circuit breaker Pattern · Security & reliability
Temporarily stop calling a failing dependency to limit repeated failures and allow recovery.
- Compensating transaction Concept · Security & reliability
An explicit business operation that counteracts a completed action when rollback is unavailable.
- Confused deputy Concept · Security & reliability
A privileged component is manipulated into acting for a party that lacks the relevant authorization.
- Data exfiltration Concept · Security & reliability
Unauthorized movement or disclosure of information through outputs, tools, or network channels.
- Excessive agency Concept · Security & reliability
Give an agent more functionality, permissions, or autonomy than the task requires.
- Goal hijacking Concept · Security & reliability
Redirect an agent from its authorized objective toward another objective.
- Guardrail Concept · Security & reliability
A check or constraint on input, output, or actions; its strength depends on where and how it is enforced.
- Indirect prompt injection Concept · Security & reliability
A redirection attempt carried in external content, such as retrieved pages, documents, or tool results.
- Jailbreak Concept · Security & reliability
An attempt to bypass a model's behavioral restrictions.
- Memory poisoning Concept · Security & reliability
Corrupt persistent information so it influences later model behavior.
- Red teaming Concept · Security & reliability
Systematically search for security, safety, and misuse failures within an authorized scope.
- Replay Concept · Security & reliability
Reconstruct or resume execution from previously recorded state or events.
- Retry policy Concept · Security & reliability
Rules for whether, when, and how often failed operations should be attempted again.
- Reward hacking Concept · Security & reliability
Optimize a reward signal in a way that defeats its intended purpose.
- Sandbox escape Concept · Security & reliability
Circumvent an execution environment's intended isolation boundary.
- Specification gaming Concept · Security & reliability
Satisfy literal measured criteria while violating the intended objective.
- Timeout Concept · Security & reliability
A limit after which an operation is considered expired or interrupted according to its execution contract.
- Tool poisoning Concept · Security & reliability
Malicious or misleading instructions placed in tool descriptions, metadata, or related tool-facing content.
Browse by topic
- Agents & architecture 23 — Who decides, what executes, and where control lives.
- Goals, specs & plans 23 — Describe the outcome before delegating the implementation.
- Loops, critics & adversaries 33 — Understand how work is challenged, repaired, and stopped.
- Multi-agent coordination 21 — Delegation, ownership, context boundaries, and aggregation.
- Context & memory 29 — What the model sees now, and what persists for later.
- Tools & output contracts 20 — Model proposals become validated, authorized operations.
- Protocols & interoperability 19 — Name the boundary: tools, agents, editors, or interfaces.
- Coding-agent internals 25 — Instructions, skills, hooks, tools, and durable artifacts.
- Retrieval & knowledge 31 — Find evidence, rank it, and preserve source boundaries.
- Evals & observability 29 — Measure outcomes and inspect the execution path.
- Inference & performance 28 — Latency, throughput, compute, and memory are different constraints.
- Models, reasoning & training 36 — Separate weight changes from context and inference-time work.
- Security & reliability 27 — Make privileges explicit and side effects recoverable.
- Generative UI & voice 18 — The interaction layer has its own contracts and timing.
- Generative & physical AI 8 — Broader model families beyond text-based assistants.
- Tools & ecosystem 44 — Recognize the role before choosing the dependency.