Glossary
Definitions of the terms used across this site — MCP, agents, webhooks, idempotency, and the tooling that surrounds them.
- ACMEThe protocol Let's Encrypt uses to issue and renew TLS certificates automatically.
- AEOAnswer Engine Optimization — optimising content to be cited by AI assistants and answer engines, not just ranked in search.
- agentAn LLM-driven program that plans, calls tools, observes results, and loops until it reaches a goal.
- agent executionA single run of an LLM agent — from initial prompt through tool calls until the agent emits a final answer or stops.
- Agent SDKAnthropic's Python and TypeScript SDK for building Claude agents that loop with tools, files, and MCP servers.
- AhrefsLeading SEO and backlink-analysis platform — keyword research, rank tracking, content gap analysis, and AI-search modules.
- AI crawlerWeb crawler operated by an AI company — GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended.
- AI ModeGoogle Search's dedicated AI tab launched in 2025 — a Perplexity-style conversational search interface inside Google.
- AI OverviewsGoogle Search's AI-generated answer block at the top of results pages, sourced from indexed structured data.
- AI SEOUmbrella term for optimising content to appear in AI-generated answers — overlaps with AEO, GEO, and LLM optimization.
- AI SEO MCPMCP server bundling AI-search optimisation tools — page audits, schema generation, robots.txt and sitemap checks.
- AirtableSpreadsheet-database hybrid with views, automations, and a REST API — used for lightweight ops databases.
- AnthropicThe AI safety company behind the Claude model family, Claude Code, the MCP protocol, and the Agent SDK.
- APIA formal contract that lets one software system call functions or fetch data from another, usually over HTTP.
- AthenaHQGEO analytics platform tracking brand visibility and citation share across major LLM answer engines.
- AtlassianThe Australian software company behind Jira, Confluence, Bitbucket, and Trello.
- bearer tokenAn API credential passed in the HTTP Authorization header — the holder is presumed authorised, no further proof.
- BlueskyOpen, federated microblogging platform on the AT Protocol — the most active X (Twitter) alternative for the tech audience.
- BreadcrumbListSchema.org type representing the trail of parent pages above the current page — drives breadcrumb display in SERPs.
- BuzzStreamOutreach CRM for link-building and PR — tracks contacts, conversations, and link-acquisition pipelines.
- CaddyModern reverse proxy and web server with automatic HTTPS via Let's Encrypt — a simpler alternative to nginx.
- canonical URLThe single URL a page declares as the official version when duplicates exist — set via a rel=canonical link.
- CDNA globally distributed cache that serves static assets from a point of presence near each user.
- ChatGPTOpenAI's consumer LLM product — chat, browse, and run plugins on the web via GPT-class models.
- CI/CDContinuous integration and continuous delivery — automated build, test, and deploy pipelines triggered on commit.
- Citation Intelligence MCPMCP server for tracking and analysing how often a domain is cited across LLM answer engines.
- Claude CodeAnthropic's official terminal-based coding agent, with MCP support and a plugin ecosystem.
- Claude Code hookUser-defined shell command triggered by Claude Code at specific lifecycle moments — session start, prompt submit, stop, and others.
- Claude DesktopAnthropic's native chat app for macOS and Windows, the original surface for installing local MCP servers.
- Claude HaikuAnthropic's smallest, fastest, cheapest Claude tier — used for bulk classification, routing, and trivial agent steps.
- Claude OpusAnthropic's most capable Claude tier — highest reasoning quality at the highest cost and latency.
- Claude SkillA markdown bundle that teaches Claude Code or Claude Desktop a domain-specific workflow on demand.
- Claude SonnetAnthropic's mid-tier Claude model family — the default workhorse balancing speed, cost, and reasoning quality.
- ClaudeBotAnthropic's crawler for both Claude training data and Claude's live web-fetch tool.
- CLIA text-only interface where a user runs a program by typing its name and arguments into a shell.
- ClineAn open-source agentic coding assistant that runs as a VS Code extension and orchestrates file edits and shell commands.
- CodexOpenAI's coding-focused product line — currently a CLI agent and IDE integration competing with Claude Code.
- Conditional AccessMicrosoft Entra ID policies that gate sign-in based on user, device, location, app risk, and authentication strength.
- Content Distribution MCPMCP server that distributes a published article to LinkedIn, Bluesky, Reddit, Hashnode, and other secondary channels.
- context windowThe maximum number of tokens an LLM can consider at once — input prompt plus output combined.
- ConvertMateE-commerce-focused SEO tool — auto-optimises product page titles, descriptions, and structured data for Shopify stores.
- CrewAIPython framework for multi-agent collaboration — define roles, assign tasks, run hierarchical or sequential crews.
- CRUDCreate, Read, Update, Delete — the four basic operations every persistent-data API exposes.
- CrunchbaseDatabase of companies, funding rounds, and executives — a primary source for B2B prospecting and competitive research.
- CursorAn AI-first code editor forked from VS Code that integrates LLM completions, chat, and agent loops natively.
- DBDatabase — any system that stores structured data and supports queries and updates over it.
- DigitalOceanDeveloper-focused cloud provider — simple Droplets (VPS), managed Postgres/Redis, App Platform, and Spaces object storage.
- DLLWindows's shared library format — a .dll file is code loaded into a process at runtime.
- DockerThe dominant container runtime — packages an app with its dependencies into an image that runs the same anywhere.
- Docker ComposeTool for defining and running multi-container Docker applications from a single YAML file.
- Docker DesktopThe official Docker GUI for macOS, Windows, and Linux — bundles the engine, CLI, Compose, and a Kubernetes node.
- DropboxIndependent cloud file storage and sync service with a stable REST API used by many automation platforms.
- DuckDBEmbedded analytical SQL database — like SQLite for columnar OLAP queries over Parquet and CSV.
- ECONNREFUSEDTCP connection-refused error — nothing is listening on the target host and port the client tried to reach.
- embeddingsDense vector representations of text — the numeric substrate for semantic search and retrieval-augmented generation.
- ETIMEDOUTNetwork timeout error — the remote host accepted the connection but did not respond within the client's timeout window.
- ETLExtract, Transform, Load — the classic data-pipeline pattern for moving data between systems.
- exponential backoffRetry strategy where each successive attempt waits exponentially longer — the standard response to rate-limit and 5xx errors.
- FAQPageSchema.org type describing a list of question-answer pairs — eligible for FAQ-rich-result treatment by search engines.
- fine-tuningContinuing the training of a base LLM on domain-specific examples to adapt its outputs to a narrower task.
- FlaskMinimal Python web framework — the standard pick for small HTTP services, internal tools, and webhook receivers.
- FraseAI-assisted content-optimization tool — SERP analysis, brief generation, and on-page scoring against ranking competitors.
- GeminiGoogle's flagship multimodal LLM family powering the Gemini app, Vertex AI, AI Overviews, and Search.
- GEOGenerative Engine Optimization — the SEO of the LLM era, focused on being surfaced inside AI-generated answers.
- Git BashA POSIX-style bash shell for Windows bundled with Git for Windows — the default Unix-ish terminal on Windows dev machines.
- GitHubMicrosoft-owned Git hosting platform — the default home for open-source code, issues, pull requests, and CI.
- Google SheetsGoogle's cloud spreadsheet — the most common lightweight database used in low-code automation.
- GPTBotOpenAI's web crawler for future model training data — does not affect ChatGPT live retrieval.
- hallucinationAn LLM output that is fluent and confident but factually wrong, including made-up citations, library functions, or quotes.
- HashnodeDeveloper-focused blogging platform with custom domains, markdown editor, and a clean API for programmatic publishing.
- HTTP RequestA generic API call — the catch-all node in automation tools when no native integration exists for a service.
- HubSpotAll-in-one CRM and marketing platform with deep webhook and REST API support for sales and lifecycle automation.
- Hugging FaceThe de facto hub for open-source ML models, datasets, and Spaces demos — the GitHub of machine learning.
- IDEAn editor bundled with project tooling — file tree, terminal, debugger, language server — for writing software.
- idempotencyA property of an operation where calling it twice has the same effect as calling it once.
- IPCInter-process communication — how processes on the same machine exchange data (pipes, sockets, shared memory).
- JiraAtlassian's project-tracking tool — the dominant issue tracker in enterprise engineering organisations.
- JSON SchemaA JSON-based vocabulary for describing the shape, types, and constraints of JSON documents.
- JSON-LDJSON for Linking Data — the JSON syntax for embedding schema.org structured data in a web page.
- JSONLNewline-delimited JSON — one JSON object per line, used for streaming logs, datasets, and LLM batch input.
- JupyterNotebook environment that interleaves code, prose, charts, and outputs — the standard interface for data science.
- KeychainApple's encrypted credential store on macOS and iOS — where Safari passwords, app secrets, and certificates live.
- LangChainPython and JavaScript framework for building LLM applications — chains, agents, memory, retrieval, and tool use.
- LangGraphLangChain's stateful-graph framework for building agents as explicit nodes, edges, and conditional branches.
- LexicalMeta's open-source rich-text editor framework — used by Facebook's posts editor and as Ghost's native post format.
- LinkedInMicrosoft-owned professional network — a key B2B distribution surface and lead source for SaaS and agencies.
- LLMLarge Language Model — a transformer-based model trained on internet-scale text to generate and reason.
- llms.txtPlain Markdown file at /llms.txt that gives AI crawlers a curated index of a site's content.
- LSPProtocol letting any editor talk to any language's analysis backend — defines completion, hover, go-to-definition.
- Make.comSaaS workflow automation platform (formerly Integromat), 1500+ integrations, visual scenario builder.
- MarkdownPlain-text markup with simple conventions for headings, lists, links, and code — the universal docs and chat format.
- MCPAnthropic's open standard for connecting LLMs to external tools and data sources over JSON-RPC.
- MediumMembership-based blogging platform — a common secondary distribution channel for tech content, with declining SEO authority.
- MetronyxAnother GEO analytics entrant — tracks LLM citations and brand share-of-voice across answer engines.
- MFAMulti-factor authentication — a second verification step (TOTP, hardware key, push) on top of a password.
- Microsoft Entra IDMicrosoft's cloud identity provider (formerly Azure AD) — the directory behind Microsoft 365 and Azure SSO.
- MistralA Paris-based AI lab shipping open-weight and commercial LLMs aimed at on-prem and EU compliance use cases.
- MSIWindows's native installer-package format — the .msi file consumed by the Windows Installer service.
- n8nOpen-source workflow automation tool with 400+ integrations, self-hostable, fair-code licensed.
- n8n-mcpMCP server that introspects an n8n instance — lists nodes, executions, and workflows for LLM-assisted workflow building.
- NotionBlock-based workspace combining documents, databases, and views — heavily used as a project hub and internal CMS.
- Notion MCPNotion's official MCP server — lets Claude and other MCP clients read and write Notion pages, databases, and comments.
- NTFSWindows's native filesystem — case-insensitive, journaled, with ACL-based permissions and alternate data streams.
- OAI-SearchBotOpenAI's retrieval crawler that fetches pages for ChatGPT search — block it and you're invisible to ChatGPT.
- OAuthIndustry-standard delegation protocol — lets a user authorise an app to access their data without sharing a password.
- OneDriveMicrosoft's consumer and business file-sync service — the storage layer behind SharePoint and Microsoft 365 Files.
- OOMOut of Memory — what happens when a process exceeds its allocated RAM and the OS kills it.
- Open GraphMeta-tag protocol that controls how a page renders when shared on social platforms — title, description, image.
- OpenAIThe AI lab behind ChatGPT and the GPT model family; Anthropic's main commercial competitor.
- Otterly.AIAI search monitoring tool — tracks brand and competitor visibility across ChatGPT, Perplexity, and AI Overviews.
- Parallels DesktopmacOS virtualisation app that runs Windows and Linux VMs natively on Apple Silicon and Intel Macs.
- ParquetOpen columnar file format optimised for analytical workloads — the standard for data-lake storage.
- People Also AskGoogle SERP feature showing related expandable questions — a major source of long-tail SEO traffic.
- PerplexityAI answer engine that searches the live web and returns synthesized answers with inline citations.
- PerplexityBotPerplexity's web crawler — the user agent that fetches pages to cite in Perplexity AI answers.
- PipedriveSales-focused CRM with a pipeline-first UI — a common alternative to HubSpot for small teams.
- PixelmojoAI search visibility and brand tracking platform competing in the same GEO analytics category as Profound and Otterly.
- PostgresThe open-source relational database — the default choice for production SQL workloads.
- PostmanCross-platform API client — build, send, and document HTTP requests interactively without writing code.
- Power AutomateMicrosoft's low-code workflow tool inside the Power Platform — automates Microsoft 365 and connected SaaS apps.
- PowerShellMicrosoft's object-oriented shell and scripting language — default on Windows, available cross-platform as pwsh.
- ProfoundAI visibility tracking platform — monitors how a brand appears in ChatGPT, Perplexity, Claude, and Gemini answers.
- prompt cachingProvider-side cache of repeated prompt prefixes — dramatically lowers cost and latency for long, stable system prompts.
- PydanticPython's de facto data-validation library — typed models that parse and validate JSON, env vars, and API payloads.
- QuoraLong-running Q&A platform — a meaningful source of long-tail SEO traffic and a frequent LLM citation source.
- RAGRetrieval-Augmented Generation — fetch relevant documents at query time and feed them to the LLM as context.
- rate limitA cap on how many requests a client can make to an API per unit of time, after which further calls are rejected.
- RedditCommunity-driven discussion platform organised into subreddits — a major source of long-tail SEO traffic and AI citations.
- RedisIn-memory key-value store — the standard pick for caches, queues, rate-limit counters, and pub/sub.
- RESTAn architectural style for HTTP APIs — resources at URLs, manipulated via standard methods like GET and POST.
- REST HookA webhook subscription pattern where the consumer registers a callback URL via REST and the producer POSTs events to it.
- retryRe-attempting a failed operation, usually with exponential backoff and a maximum attempt count.
- Rich ResultsGoogle's enhanced SERP listings — recipe cards, review stars, FAQ accordions, product info — powered by Schema.org markup.
- robots.txtPlain text file at /robots.txt that tells crawlers which paths they may or may not fetch.
- RPCRemote Procedure Call — a calling convention where invoking a remote function looks like a local call.
- SaaSSoftware delivered as a multi-tenant web service on a subscription — the dominant business model for B2B software.
- schema.orgOpen vocabulary for structured data on the web, co-maintained by Google, Microsoft, Yahoo, and Yandex.
- SDKA library plus tooling that wraps a service's API in a specific programming language.
- SE RankingAll-in-one SEO platform with keyword tracking, site audit, backlink monitoring, and AI Overview/LLM citation tracking.
- selectorA pattern that targets HTML elements — used by CSS for styling and by scrapers for extraction.
- SERPSearch Engine Results Page — what Google or Bing shows for a query, including organic, ads, AI Overviews, and PAA boxes.
- SharePointMicrosoft's collaboration and document-management platform tightly integrated with Microsoft 365 and OneDrive.
- sitemapAn XML file listing the URLs a site wants search engines and AI crawlers to discover and index.
- SlackSalesforce-owned team chat platform with a deep app ecosystem — the most common destination for automation notifications.
- SQLiteEmbedded, file-based SQL database — the most-deployed database in the world, runs without a server process.
- SSEServer-Sent Events — a one-way HTTP streaming protocol used for LLM token streaming and MCP HTTP transport.
- SSHEncrypted protocol for remote shell access and tunnelled file transfer — the standard way to reach a server.
- SSOSingle sign-on — log in once with your identity provider and access many apps without re-authenticating.
- StripeThe dominant payments API for SaaS — checkout, subscriptions, invoicing, and webhooks for revenue events.
- structured dataMachine-readable metadata embedded in a page that describes its entities — consumed by search and AI engines.
- system promptThe instruction text given to an LLM that sets its persona, rules, and capabilities before any user message.
- TLSThe protocol that encrypts HTTP and other connections — what the lock icon in browsers actually means.
- tokenThe unit an LLM processes — a sub-word chunk produced by the model's tokenizer, billed individually.
- tool useAn LLM feature where the model can request execution of named functions (tools) and receive results before responding.
- TTLTime-to-live — how long a cached or transient value remains valid before being evicted or re-fetched.
- UIUser interface — the visual and interactive surface a human uses to operate software.
- UiPathEnterprise RPA platform — record desktop/browser interactions and replay them as Robot processes.
- UUIDA 128-bit globally unique identifier — the standard primary-key choice for distributed systems.
- UXUser experience — the total quality of a user's interaction with a product, including UI, flow, and emotional response.
- VPSVirtual Private Server — a virtualized Linux box rented from a cloud provider, billed by the hour or month.
- VS CodeMicrosoft's free cross-platform code editor, the most widely used IDE for general-purpose development.
- WAFA reverse-proxy filter that blocks common web exploits before they reach the origin server.
- WebFetchClaude Code's built-in tool for fetching a URL and returning its text content for the model to read.
- webhookAn HTTP POST sent by one system to a URL on another system when an event occurs — push, not poll.
- WebSearchClaude Code's built-in web search tool — returns ranked result URLs and snippets for a query.
- WikidataWikipedia's structured-data sibling — a free knowledge graph queried via SPARQL, widely cited by LLMs.
- WikipediaThe free online encyclopedia — the single most-cited domain in LLM answer surfaces by a wide margin.
- WiXOpen-source toolset for authoring Windows Installer (.msi) packages from XML source files.
- WSLA real Linux kernel running alongside Windows — the standard way to develop Linux workloads on Windows.
- X (Twitter)Microblogging platform rebranded from Twitter — still a primary distribution channel for developers despite migration losses.
- YAMLHuman-readable data-serialisation format — the standard for config files in DevOps and CI/CD.
- YouTubeGoogle's video platform — a major surface for long-form tutorials and the most-cited video source in AI answers.
- ZapierSaaS workflow automation pioneer, 6000+ integrations, charges per task (each step run).
- Zapier MCPZapier's MCP server exposing thousands of Zapier-connected SaaS actions as tools to MCP clients like Claude.