Skip to content

Glossary

MCP and agent terms, in plain words

One plain definition per term. Terms with a page add an example, common confusions and links to the guides and docs that use them; the rest link to the matching L2M feature. Hover or focus a term anywhere in the docs to see its definition.

100 terms

A

Agent loop
The cycle an AI agent repeats: call the model, run the tools it asks for, add the results to the conversation, then call the model again until it answers.
Agent memory
What an agent stores from earlier turns or runs and feeds back into later model calls: conversation history, summaries, saved tool results or retrieved notes.
AI agent
A system in which a language model decides, step by step, which tools to call and in what order, reading each result before it chooses the next step, until it can answer.
Approval gate
A step that pauses a run until a person approves or rejects what happens next, such as a payment, an email or a write tool call. In L2M: Approval gate
Audit log
A record of who did what and when, such as changes to settings, rules and secrets, kept so that actions can be checked later. In L2M: Audit log

B

Breaking change (MCP tool)
A change to a server's tools that can break existing callers, such as a removed tool, a removed input property or a new required argument. In L2M: Breaking change (MCP tool)
Budget
A cap on model spend or tokens for one scope, such as a project, a user or a client, over a period, which warns or blocks new work when it is reached.

C

Cassette
A recording of the outside calls a run made, with their answers, so a later test run can be answered from it instead of calling the real services. In L2M: Cassette
Confused deputy
A program that holds more authority than its caller and is tricked into using that authority on the caller's behalf, doing something the caller could not do directly.
Context window
The most text, counted in tokens, that a model can take into account in one call: system prompt, conversation, tool definitions and tool results, plus the answer it writes.

D

Data exfiltration
Data leaving a system without permission. Through an agent it can leave in a tool call, a link, an image URL or a webhook the model was steered to use. In L2M: Data exfiltration
Data labels
Tags that travel with data through a workflow or agent, recording where it came from and what it contains, so later checks can decide what that data may do.
Data retention
How long each kind of stored data is kept before it is deleted, such as run history, tool-call logs and agent memory. In L2M: Data retention
Data-subject request
A person's request to an organization about their own personal data, such as a copy of it or its deletion, under privacy laws like the GDPR. In L2M: Data-subject request
Delegated credentials
Each person's own account is used for a tool call, instead of one shared service credential, so the server sees who is acting and applies their permissions. In L2M: Delegated credentials

E

Elicitation (MCP)
An MCP server asking the person for input through the client, usually while it handles a request such as a tool call: a message with a simple form, or a URL to open for sensitive steps. In L2M: Elicitation (MCP)
Eval
A test for an AI system: give it an input, run it, and apply grading logic to the result to decide whether it succeeded.
Excessive agency
OWASP's name for an LLM application that can take damaging actions because it has more functions, permissions or autonomy than its task needs.

G

Golden run
A recorded run of a workflow or agent that you have checked and keep as the reference: its input, every step's result and every outside call, for replaying changes against.
Golden set
A curated collection of reference cases, each an input with a known-good result, that a new version of an AI workflow or agent must pass before it ships.
Grader
The logic that scores one result in an eval: an exact comparison, a schema check, a script, a model acting as judge, or a person.
Grounding
Keeping a model's answer to what the documents it was given actually support, so each claim can be traced back to a source.

H

Human in the loop
A design in which a person can see, approve, change or stop an AI system's action before it takes effect, instead of the system acting alone.

I

Improper output handling
OWASP LLM05: passing a model's answer to another system (a browser, a shell, a database) without validating it first, so the answer can act as an attack.
Indirect prompt injection
Prompt injection in which the attacker's instructions hide in content the model reads while doing its job, such as a web page, an email, a document or a tool result.

J

Jailbreak
Input written to make a model ignore its safety rules. OWASP treats it as a form of prompt injection; prompt injection more broadly hijacks an application's instructions.

L

Least privilege
The security principle that each user, program or agent gets only the access it needs for its task, and nothing more.
Legal hold
An instruction to keep certain data from being deleted, whatever its age, because it may be needed for a legal matter. In L2M: Legal hold
Lethal trifecta
Simon Willison's name for an agent that combines access to private data, exposure to untrusted content and a way to send data out, which lets one prompt injection steal data.
LLM as judge
Using a language model to grade another model's output against a written rubric, for qualities that exact comparisons cannot check. In L2M: LLM as judge
LLM gateway
A service between applications and model providers that gives them one API endpoint and adds per-client keys, routing, spend limits, rate limits and checks on prompts and answers.

M

Masking
Replacing a sensitive value, such as an email address or a card number, with a partly hidden form before the text reaches a model, a log or a person.
Master key
The one key that encrypts the other stored secrets. Without it the stored secrets cannot be decrypted, so it is backed up apart from the database. In L2M: Master key
Max iterations
The most rounds of model call and tool calls an agent may take in one run. When it is reached, the agent stops instead of looping on. In L2M: Max iterations
MCP client
The part of an AI application that holds one connection to one MCP server: it negotiates the protocol version and capabilities, then sends that server's requests and receives its replies.
MCP gateway
A service that sits between MCP clients and MCP tools and serves them through one endpoint, adding per-client credentials, access rules, limits and a log of every call.
MCP host
The AI application, such as a chat app, an IDE or an agent runtime, that creates MCP clients, connects them to servers and decides what the model may use. In L2M: MCP host
MCP prompt
A reusable message template an MCP server offers, with named arguments, for a person or application to pick, rather than for the model to call. In L2M: MCP prompt
MCP resource
Data an MCP server shares by URI, such as a file or a record, for the application to read. Unlike a tool, it is usually the application, not the model, that decides when to read it. In L2M: MCP resource
MCP server
A program that offers tools, resources or prompts to AI applications over the Model Context Protocol, locally over stdio or remotely over HTTP.
MCP session
Over Streamable HTTP, the state that ties a client's requests together: if the server returns an MCP-Session-Id when the client initializes, the client sends it with every later request. In L2M: MCP session
MCP tool
A named action an MCP server offers, described by a JSON Schema for its input, that a model can ask to call and a client runs with a tools/call request.
MCP tool policy
A set of rules, enforced outside the model, that decides whether a given agent or client may call a given MCP tool with given arguments: allow it, deny it, or hold it for a person to approve.
MFA
Multi-factor authentication: signing in with a second proof, such as a one-time code from an authenticator app, as well as a password or SSO. In L2M: MFA
Misinformation (LLM)
OWASP LLM09: a model stating false or unsupported things with confidence, and people or systems acting on them without checking.
Model allowlist
A list of the providers and models an organization permits. Calls to any other model are refused. In L2M: Model allowlist
Model Context Protocol (MCP)
An open protocol that lets AI applications discover and use tools, data and prompt templates from separate servers through one standard set of JSON-RPC messages.

O

OAuth 2.1
A draft consolidation of OAuth 2.0 and its security best practices, such as PKCE for every client. MCP authorization for HTTP servers builds on it. In L2M: OAuth 2.1
OIDC
OpenID Connect: an identity layer on top of OAuth 2.0 that tells an app who signed in, using signed ID tokens from an identity provider. In L2M: OIDC
OpenAI-compatible API
An HTTP API that accepts the same requests as OpenAI's API (such as /v1/chat/completions), so any client built for it can switch servers by changing the base URL. In L2M: OpenAI-compatible API
OpenTelemetry
An open standard and set of tools for traces, metrics and logs, sent over its own protocol (OTLP) to any compatible backend. In L2M: OpenTelemetry
Output checks
Tests run on a model's answer before anything uses it: whether it parses and fits a schema, stays on allowed topics, avoids unsafe content and sticks to its sources.

P

pass@k and pass^k
Two ways to score k tries of the same test: pass@k is the chance that at least one try succeeds, pass^k the chance that all k succeed.
PII
Personally identifiable information: data that identifies a person alone or combined with other data, such as a name, an email address or an ID number.
PKCE
Proof Key for Code Exchange: the client sends a hash (S256) of a one-time secret with the authorization request and the secret itself when it redeems the code, so a stolen code is useless. MCP clients must use it.
Production capture
Recording a sample of real production runs, with personal data masked, so the useful ones can become test cases. In L2M: Production capture
Prompt injection
Input that makes a language model ignore or change the instructions its application gave it, so it does something the application did not intend.
Protection token
A placeholder that stands in for a sensitive value in a prompt or a log. A reversible token can be turned back into the value later, only where a rule allows it. In L2M: Protection token

R

RAG (retrieval-augmented generation)
A pattern that first retrieves documents relevant to a question and puts them in the prompt, so the model answers from that text instead of from memory alone.
Rate limit
A cap on how many requests a caller may make in a period, so one client cannot flood a service or run up its costs. In L2M: Rate limit
RBAC
Role-based access control: permissions are given to roles, such as admin or viewer, and people get roles, rather than permissions one by one. In L2M: RBAC
Regression eval
An eval that checks a system still does what it did before a change. It should pass nearly every time; a failure means something broke.
Release gate
A check in the release path that lets a new version of a workflow or agent go live only after its tests passed on that exact version.
Replay
Running a changed workflow or agent on the input of a recorded run, with its calls to the outside world answered from the recording, and comparing the result with the original.
Right to erasure
The GDPR right (Article 17) to have one's personal data deleted on certain grounds, with exceptions, for example data needed for legal claims. In L2M: Right to erasure
Rug pull
An attack in which an MCP server changes a tool's definition after people reviewed and approved it, so a tool trusted as harmless becomes harmful without a new review.

S

SAML
Security Assertion Markup Language: an older XML-based standard for single sign-on, where the identity provider sends a signed assertion about the user. In L2M: SAML
Sampling (MCP)
An MCP server asking the client to run a model call on its behalf, so a tool can use the client's model without its own API key. In L2M: Sampling (MCP)
SBOM
Software bill of materials: a list of every component and version inside a piece of software, used to check it for known vulnerabilities and licences. In L2M: SBOM
Secret reference
A pointer to a stored, encrypted secret that a workflow or config holds in place of the secret itself, so the raw value never appears in exports or responses. In L2M: Secret reference
Self-hosted
Software that you run on your own servers or cloud account, so its data and its keys stay under your control. In L2M: Self-hosted
Sensitive information disclosure
OWASP LLM02: a model or app revealing personal data, credentials or confidential business data in its answers, logs or tool calls.
Service account
An account for a program rather than a person, such as a CI job, with its own keys and its own limited permissions. In L2M: Service account
Session tool cache
A store of full tool results kept outside the prompt for one conversation, so the agent can fetch an earlier result again instead of calling the tool twice. In L2M: Session tool cache
Shadow mode
Running a new rule on real traffic and recording what it would have done, without enforcing it, so you can check it before it blocks anything. In L2M: Shadow mode
Signed run manifest
A record of what one run did, signed with a private key, so anyone with the public key can check later that it was not changed. In L2M: Signed run manifest
Snapshot mock
A stand-in MCP server built from a real server's recorded tool schemas, so tests can run without the real server or a network. In L2M: Snapshot mock
Spotlighting
A family of prompt techniques that mark which text came from outside sources, so the model treats it as data to work on rather than instructions to follow.
SSO
Single sign-on: people sign in once with their organization's identity provider and reach many apps without a separate password for each. In L2M: SSO
SSRF
Server-side request forgery: making a server fetch a URL the attacker chose, such as an internal service or a cloud metadata address, from inside the network. In L2M: SSRF
stdio transport
The MCP transport in which the client starts the server as a child process and exchanges newline-delimited JSON-RPC messages over the process's standard input and output.
Streamable HTTP
The MCP transport for servers reached over a network: every client message is an HTTP POST to one endpoint, and the server answers with a JSON body or a server-sent events stream.
Structured tool output
A tool result returned as JSON in structuredContent, besides the text content. A tool that declares an outputSchema must return results that match it. In L2M: Structured tool output
Subject reference
A value saved with a run that identifies the person the run was about, so that person's runs can be found later without searching every record. In L2M: Subject reference
Supervisor agent
An agent whose tools include other agents: it splits a task, hands each part to a worker agent, and combines the answers into one reply.
System prompt leakage
OWASP LLM07: the hidden instructions of an app getting out, which matters most when they hold secrets or describe access rules that should be enforced elsewhere.

T

Token (LLM)
The unit a language model reads and writes: a word, part of a word or a punctuation mark. Context limits, prices and rate limits are all counted in tokens.
Token passthrough
An anti-pattern in which an MCP server accepts an access token that was not issued for it and forwards that token to a downstream API; the MCP specification forbids it.
Tool annotations
Optional hints an MCP server attaches to a tool definition, saying whether the tool is read-only, destructive, idempotent or reaches the outside world; clients must treat them as untrusted.
Tool calling
A model's structured request to run one named tool with JSON arguments, instead of answering in text. The application runs the tool and returns the result.
Tool poisoning
An attack that hides instructions for the model in a tool's description, schema or results, so the agent follows them while the user sees only a harmless tool.
Tool shadowing
One MCP server's tool descriptions changing how an agent uses another server's tools, for example by telling it to copy data into every email it sends.
Tool-call telemetry
A record of every tool call an agent or client makes: which tool, how long it took, whether it failed, and what it cost. In L2M: Tool-call telemetry
Toolset
A named, curated group of tools, drawn from several MCP servers and workflows, served at one MCP endpoint so each client sees only the tools it needs.

U

Unbounded consumption
OWASP LLM10: an application that lets users or agents run up model calls, tokens or compute without limits, which costs money and can deny service to others. In L2M: Unbounded consumption

V

Vector store
A database that keeps embeddings (lists of numbers that stand for the meaning of text chunks) and returns the chunks whose embeddings are nearest to a query's.

W

Worker agent
A specialist agent, with its own model, tools and instructions, that a supervisor agent calls like a tool for one part of a task. In L2M: Worker agent
Workflow DAG
A workflow drawn as a directed acyclic graph: steps joined by edges that never loop back, run in the order their inputs allow. In L2M: Workflow DAG
Write tool
A tool that can change something outside the agent, such as a file, a record or a message, as opposed to a read-only tool that only fetches data. In L2M: Write tool

This website (not the L2M product) uses Google Analytics, only after you accept cookies. Cookie settings