AI Coding Dictionary
The vocabulary of AI coding, translated into plain English.
AI coding can feel like it is just for experts. The basic terms of engagement are learnable in an afternoon. Once you have them, the whole thing stops feeling like guesswork.
Plain-English reference
Skimmable definitions for the terms that make AI coding click.
Section 1 — The Model
14 terms- ModelThe parameters. Stateless — does nexttoken prediction and nothing else. "Claude Opus 4.7" and "GPT5" are models. On its own a model can't do anything agentic; it has to be harne...
- ParametersThe numbers inside a model — often billions of them — tuned during training. Everything the model "knows" lives in them. Training sets them; inference uses them unchanged. Also...
- TrainingThe process that sets a model's parameters, by exposing it to vast amounts of text and adjusting parameters to improve nexttoken prediction. A onetime, expensive process done by...
- InferenceRunning a trained model to generate output — what happens on every model provider request. Parameters stay fixed; the model just does nexttoken prediction over the context it's...
- TokenThe atomic unit a model reads and writes. Roughly wordsized but not exactly — common words are one token, rare or long ones split into several. Context window size, cost, and la...
- Next-token predictionWhat the model actually does. Given a context, it samples one next token, appends it, and runs again. Every output — a sentence, a tool call, a thousandline file — is built one...
- Non-determinismThe same input can produce different output. Run a model twice with identical context and you may get two different answers — sometimes a word, sometimes a completely different...
- Model providerWhatever serves a model for inference. Usually a remote service (Anthropic, OpenAI, Google), but can also be local — Ollama, LM Studio, llama.cpp running on your own machine. Th...
- HarnessEverything around the model that turns it into an agent: tools, system prompt, contextwindow management, permissions, hooks. Claude.ai and Claude Code run on the same model but...
- Model provider requestOne roundtrip from the harness to the model provider. The harness sends the current context; the provider returns one response (a tool call or a final answer). A single user mes...
- Input tokensTokens the harness sends on each model provider request. Billed at a lower rate than output tokens.
- Output tokensTokens the model generates back. Billed at a higher rate than input tokens, since they cost more compute to produce.
- Prefix cacheThe providerside store that lets consecutive model provider requests skip reprocessing a shared prefix. When the start of a request matches the start of a recent one — same syst...
- Cache tokensInput tokens the provider has cached from a previous model provider request so it doesn't have to reprocess them. When consecutive requests share a prefix, the provider reuses t...
Section 2 — Sessions, Context Windows & Turns
8 terms- StatelessCarries no information forward. The model is stateless across model provider requests — each request resends the full context window, because the model has no way to see anythin...
- ContextThe relevant information the agent has access to right now. The abstract noun — not the raw input the model sees (that's the context window), not the running history (that's the...
- Context windowEverything the model sees on each model provider request. Finite, modelspecific, and the only surface through which the model perceives anything.
- StatefulCarries information forward. A session is stateful across turns — context accumulates as the session runs, which is why long sessions drift into the dumb zone. An agent can be m...
- AgentA model harnessed with tools, a system prompt, and a context window, that takes turns with a user. Claude Code is an agent. Cursor is an agent. Claude.ai is an agent. An agent i...
- System promptThe instructions the harness prepends to every model provider request — the agent's standing brief: who it is, how to behave, which tools it can call, what conventions to follow...
- SessionOne bounded run of interaction with an agent. Starts empty, accumulates messages, tool results, and files read, and ends when cleared, closed, or compacted into a fresh session....
- TurnOne user message plus everything the agent does in response, up until it yields back to the user. Contains one or more model provider requests — many, if the agent calls tools....
Section 3 — Tools & Environment
9 terms- EnvironmentThe world the agent acts on — anything outside the harness that the agent perceives through tool results and changes through tool calls. The harness runs the agent; the environm...
- FilesystemA tree of files and directories the agent reads from, writes to, and executes within — the default kind of environment for a coding agent. AGENTS.md, skills, source code, build...
- ToolA function the harness exposes for the agent to call — Read, Write, Bash, Search. Tools are how an agent perceives and acts on the environment: it can't see the environment exce...
- Tool callThe model's output naming a tool and its arguments — just structured text. It doesn't do anything on its own; the harness has to read it and execute. Produced by the model in on...
- Tool resultWhat the harness sends back after executing a tool call — the file contents, the command output, the error. The agent's only window onto the environment. Travels back to the mod...
- Permission requestWhat the harness shows the user before executing a tool call that isn't preapproved. The model produces a tool call; instead of running it immediately, the harness pauses and as...
- Permission modeThe permissiongating slice of an agent mode — which tool calls trigger a permission request and which run automatically. The original purpose of mode systems before harnesses st...
- Agent modeA preset that shapes how the agent operates at runtime — bundles a permission mode with behavioral instructions injected into the system prompt. Examples: a default that prompts...
- SandboxAn isolated environment the agent runs inside — a container, VM, ephemeral filesystem, or restrictedpermission shell. Limits the blast radius of agent actions: even if the agent...
Section 4 — Failure Modes
9 terms- SycophancyConfidently agreeable model output. Caused by training: the model was shaped to favor answers humans liked, and humans tend to like agreement more than they like being told they...
- HallucinationConfidentlywrong model output. Two flavors with different causes and fixes:
- Parametric knowledgeWhat the model "knows" from training, stored in its parameters. Frozen at training time — the model can't see its own parameters or update them. Detail is lost in the squeeze: b...
- Knowledge cutoffThe date past which a model has no parametric knowledge. Libraries, APIs, and events from after the cutoff are fabrication traps unless their docs are loaded as contextual knowl...
- Contextual knowledgeFacts the agent can read directly from the context right now — the user's task, files the agent has read in, tool results, AGENTS.md content loaded at session start. Counterpart...
- Attention relationshipWhen predicting each token, the model factors in every other token in the context — some heavily, others barely at all. The pairing between two tokens is an attention relationsh...
- Attention budgetEach token has a finite amount of influence to distribute across the rest of the context. Heavy influence on one relationship leaves less for others. The budget is pertoken and...
- Attention degradationAs a session grows, each token's attention budget is spread across more competitors. The signal on any one meaningful relationship shrinks; noise from irrelevant context crowds...
- Smart zoneEarly in a session the agent is in a "smart zone" — sharp, focused, recall is good. As the session grows it drifts into a "dumb zone": sloppier, forgetful, more mistakes — and m...
Section 5 — Handoffs
7 terms- ClearingEnding the current session and starting a fresh one. The next message begins with an empty session and an empty context window. Usually userdriven.
- HandoffTransferring agent context from one session to another, with no return path. The carry mechanism varies — a written handoff artifact, an inmemory summary (compaction), and other...
- Handoff artifactA document used as the carry mechanism for a handoff — written by one session to be read by another. One way among several (see also compaction, compaction).
- SpecA handoff artifact describing a multisession piece of work — what's being built, not how each session does its share. Mutates as work progresses. Made of tickets.
- TicketA handoff artifact scoping one session of work. Stands alone, or hangs off a spec as one of its children. Tickets can block or be blocked by sibling tickets, so the order of wor...
- CompactionA handoff done inmemory: the previous session's history is summarised and seeds a fresh session. Lossy — detail traded for headroom. Triggered manually by the user, or automatic...
- AutocompactCompaction triggered automatically by the harness when the context window approaches full.
Section 6 — Memory and Steering
5 terms- Memory systemA system that attempts to make an agent stateful across sessions. Persists information into the environment during a session and reloads it into the context window at the start...
- AGENTS.mdA file in the environment that the harness loads into the context window at session start — the project's standing brief to the agent. Crossharness convention.
- Progressive disclosureLoading only the context an agent needs right now, with pointers to the rest. Borrowed from UI design.
- SkillA teachable capability bundled as a unit — instructions and resources for doing one task well, kept in the environment and loaded into the context window only when relevant. The...
- SubagentAn agent spawned by another agent via a tool call. Runs in its own session with its own context window, and reports a single tool result back. Distinct from a handoff — the pare...
Section 7 — Patterns of Work
8 terms- Human-in-the-loopA working pattern where one or more humans pair with the agent during a session — reviewing, redirecting, or collaborating in real time. The human is present and engaged, not ju...
- AFKA working pattern where the user kicks off a session and leaves the agent to run unattended. The throughput multiplier of AI coding — many AFK sessions can run in parallel while...
- Automated checkA deterministic verification that runs in the environment — tests, type checks, lints, build, precommit hooks. Pass/fail, no judgement. The signal an agent can selfcorrect from...
- Automated reviewAn agent reviewing another agent's work, often with a different model or system prompt. Nondeterministic: it forms a judgement. Runs anywhere — premerge on a PR, posthoc on comm...
- Human reviewThe user reading the code the agent produced and forming a judgement on it. Reading the diff or the changed files counts; reading the agent's description of what it did does not...
- Vibe codingA working pattern where the user accepts the agent's code without human review. The diff is treated as opaque — what matters is whether the program behaves, not what's inside. A...
- Design conceptThe shared understanding of what's being built, held in common between user and agent but separate from any asset. Brookes' term (The Design of Design): the conversation, handof...
- GrillingA technique for developing a design concept with an agent: the agent interviews the user Socratically, one decision at a time, proposing a recommended answer for each. Slows the...