Glossary
A term marked (website term) comes from the ekline.io website. Its definition gives the app’s name for the same thing.
| Term | Definition | See |
|---|---|---|
| Agent Task journey | A journey that gives a coding agent, Claude Code or Codex, a task to do with your product, such as following your quickstart. Each run starts in an empty working folder. | Agent Task journeys |
| AI SEO and GEO (website term) | The ekline.io terms for how AI assistants such as ChatGPT and Claude describe and recommend your product. In the app, Journeys measures what they say, and the Docs Agent and the Marketing Agent fix the pages they read. | Turn a gap into a fix |
| AI visibility (website term) | The ekline.io term for whether AI assistants name and recommend your product. In the app, a Buyer questions report measures it, and a shared report’s public page is titled AI visibility report. | Read a Buyer questions report |
| Automatic PR review | The Docs Agent checks each new pull request or merge request in a repository you selected and judges how likely it is to need a docs update: Low, Medium or High confidence. Depending on that confidence, it stays silent, comments with a suggestion, or starts a docs update. | Automatic documentation review on pull requests |
| Automation | A prompt and a schedule: each run starts a Docs Agent session that carries out the prompt. Turn on a built-in one or write your own under Automations in the sidebar. | Automations |
| Baseline | A collection’s first run, which later runs of it are compared with question by question. On a journey’s reports, baseline marks the first report for each agent and model. | Collections and sharing |
| Brand names | The Company Profile field with your product’s names and the other spellings people use for them. Checks that look for your product, such as you are recommended, not just mentioned, need it. | Company Profile and Domains |
| Buyer question | What a Buyer questions report, its public page and its PDF call each journey in the collection. | Read a Buyer questions report |
| Buyer questions report | The report a collection run produces: each buyer question marked You win, Split or You lose, who was recommended instead of you, and up to three fixes to start first. Your first one comes from the free report, which creates a collection named Buyer questions. | Read a Buyer questions report |
| Check | A test each answer in a journey must pass, such as you are recommended, not just mentioned. A check is measured or judged. | Write checks that mean something |
| Citation | A URL an agent reported as a source for its answer; Codex reports none. Reports count citations by site and mark pages under the documentation sites in Company Profile as yours. | Read a Buyer questions report |
| Collection | A set of journeys that run together and share one report. Each journey keeps its own agents, models and number of runs. | Collections and sharing |
| Company Profile | Your company’s product, brand names, documentation sites, audiences, personas, competitors, and tone and voice, under Settings > Organization > Company Profile. The Marketing Agent and Journeys use it. To check whether the Docs Agent uses it too, contact support@ekline.io. Some screens call it Company Info. | Company Profile and Domains |
| Credit | The unit your organization’s EkLine usage is counted in. Agent work draws credits from one balance for your organization. | Credits and usage |
| Custom instructions | Text that tells one agent how to write for your organization, added to every new session of that agent. The Docs Agent and the Marketing Agent each have their own, up to 20,000 characters. | Custom instructions |
| Docs Agent | The agent that writes and updates your documentation from your code, pull requests, tickets and existing docs. It delivers changes as pull requests or merge requests in your docs repositories. | Agents overview |
| Docs Reviewer | EkLine’s style checker for documentation: it checks each docs change against your style guide and comments on what to fix. It runs in CI, VS Code, Google Docs or the CLI. | How EkLine Docs Reviewer works |
| Domains | Email domains your organization has claimed, so teammates who sign up with one are offered to join. Owners and Admins manage them under Settings > Organization > Domains; the agents do not read them. | Members and roles |
| Editor | The page in the EkLine app where you give an agent a task, review what it drafts, and send the result on. | Work in the editor |
| EkLine token | Your organization’s token, under Settings > Organization > Access, that you add as EK_TOKEN when you set up the Docs Reviewer. Work done with it is recorded as the organization, not a person. | Members and roles |
| Free AI report (website term) | The ekline.io offer behind your first Buyer questions report: EkLine asks 15 buyer questions 10 times each on ChatGPT, Claude, Perplexity and Gemini. In the app it is called the free report, and each organization gets one. | Get your first AI report |
| Graded answer | In a collection, an answer from a finished run whose checks produced a pass or a fail: the deciding check’s result when the question has one, otherwise any check’s. Only graded answers count toward You win, Split and You lose. | How results are measured |
| Integration | A tool you connect for your whole organization under Settings > Integrations, such as Slack, Jira or Google Drive. | Integrations overview |
| Jira trigger | A rule under Settings > Docs Agent > Jira triggers that starts a Docs Agent session when a ticket in the projects and assignees you select enters a status you select. | Start work from Jira tickets |
| Journey | A question or task to test with AI agents: one to 20 turns, a persona, the checks that grade each answer, and the agents to run it on. A journey is a Research journey or an Agent Task journey. | Journeys |
| Judged check | A check an AI model grades against a written rubric, such as the answer is correct. | Write checks that mean something |
| Knowledge base | In Settings > Integrations, the Knowledge Base tab is where you upload reference documents. KB Management is where you select the Confluence spaces and Pylon knowledge bases the Docs Agent can update. | Manage a Confluence or Pylon knowledge base, Reference documents |
| Marketing Agent | The agent that drafts blog posts, LinkedIn posts and newsletters in your company’s voice. Drafts stay in its editor for you to copy or download. | Marketing Agent |
| MCP client | An AI tool, such as Claude Code or Cursor, that connects to EkLine’s MCP server by sending your personal API key in a request header. | Connect an MCP client |
| MCP server | EkLine’s server at https://ekline.io/api/mcp. With a personal API key, an MCP client can start and follow agent sessions, find past sessions, search the EkLine docs, read or change your company profile, and add Docs Agent rules. | MCP tools |
| Measured check | A check that code decides without an AI model, such as you appear at turn 1. | Write checks that mean something |
| No result | A buyer question with no graded answers, so it is not marked You win, Split or You lose. | How results are measured |
| Not checked | A check result, drawn as a white, outlined square, for a check that could not decide: for example, Company Profile has no brand names, or the judge found too little evidence. A collection leaves these results out of its counts. | How results are measured |
| Organization memory | Notes the agents write for themselves while they work, shared by every session in your organization. You cannot view or edit them in the app. | Organization memory |
| Persona | In a journey, the person the agent acts as while it works, written by you or selected from the personas in Company Profile. | Create a research journey |
| Personal API key | A key from Settings > Profile > Personal API keys that lets an MCP client work in EkLine as you, in one organization. EkLine shows it only once. | Connect an MCP client |
| Powered by EkLine badge | A credit you can embed on your documentation site, as a web component or an SVG badge. It is not related to the credits your usage is counted in. | Add the Powered by EkLine credit |
| Published journey | A journey that has finished at least one run. Only published journeys can be added to a collection. | Collections and sharing |
| Reference document | A file you upload under Settings > Integrations > Knowledge Base, such as a product brief or glossary, that the agents can read as context. Each file can be up to 3 MB. | Reference documents and repository notes |
| Report | In Journeys, what you get each time you start a journey: every run from every agent you selected, shown as a scorecard of checks against runs. A collection run produces a Buyer questions report. | Read a Buyer questions report |
| Repository note | A short description of one repository, added on its card under Settings > Integrations > GitHub Integration or GitLab Integration. Both agents read it as a description, not an instruction, and it can be up to 5,000 characters. | Reference documents and repository notes |
| Research journey | A journey that asks an agent a question, such as which product to use for a job, and checks the answer. Any of ChatGPT, Claude, Perplexity, Gemini, Claude Code and Codex can run it. | Create a research journey |
| Review rules | Settings on one repository that decide which pull requests automatic PR review looks at, by branch, author, title, label or file. Open them from the Review rules icon on the repository in your GitHub or GitLab integration settings. | Control which pull requests EkLine reviews |
| Reviewer Rules | Settings > Docs Agent > Reviewer Rules, where you enable or disable the Docs Reviewer’s built-in style checks. They configure the Docs Reviewer, not the agents’ instructions. | EkLine rules reference |
| Role | Owner, Admin or Member: what a person can do in your organization. Owners and Admins invite members, and only the Owner manages billing. | Members and roles |
| Rules | Standing instructions under Settings > Docs Agent > Rules, copied into every new agent session. A General rule is always in force; a Review rule applies before a docs pull request is opened or updated. | Which setting to use |
| Run | In Journeys, one agent working through every turn of a journey once, as one conversation; Number of runs sets how many each agent does. For an automation, a run is each time it starts its Docs Agent session, on schedule or with Run now. | How results are measured |
| Sandbox | A site URL and sign-in variables, under Settings > Docs Agent > Sandbox, that let the Docs Agent sign in to pages behind a login, for example to capture screenshots. | Browse authenticated pages with a sandbox |
| Scheduled agent | The earlier name for an automation, still used in older changelog entries. | Automations |
| Session | One conversation with the Docs Agent or the Marketing Agent, with its own drafts. Find past sessions in the app sidebar. | Find a past session |
| Skill | A workflow your team repeats, saved under Settings > Docs Agent > Skills and run by typing its slash command, such as /release-notes-checklist. The agent also uses a skill on its own when a task matches its description. | Package a workflow as a skill |
| Slack triggers | The Settings > Docs Agent tab with one switch, Automatically raise PRs from Slack requests, which has the Docs Agent open a pull request for changes drafted from a Slack mention. It is on by default. | Docs Agent Slack bot |
| Split | A buyer question where more than 20% and less than 80% of its graded answers, across every assistant, passed. Only a collection’s report gives this result. | How results are measured |
| Support knowledge base (website term) | The ekline.io term for keeping your help center in step with your product. In the app, the Docs Agent updates Confluence pages and Pylon articles, and automations turn Pylon tickets into draft articles. | Manage a Confluence or Pylon knowledge base, Answer recurring support questions |
| Turn | In a journey, one message sent to the agent; a journey has 1 to 20, sent in order in one conversation. In a session, one message you send and the agent’s work on it, which is how credits are counted. | Create a research journey |
| Usage | The Usage page, in your account menu, which shows the credits your organization used by calendar month or by day, for Docs Agent, Marketing Agent and Journeys, and when the current cycle resets or ends. | Credits and usage |
| Visual Style Guidelines | The Company Profile field with your brand colors and style notes, which the agents follow when they draw images. | Visual style guidelines |
| What else EkLine can read | Repositories EkLine reads to work out what changed and why. It never edits files in them or pushes to them, though it can comment on their pull requests. | How EkLine opens pull requests |
| Where EkLine writes | Repositories that hold your documentation, where EkLine opens pull requests (merge requests on GitLab). | How EkLine opens pull requests |
| You lose (lost) | A buyer question where 20% or fewer of its graded answers, across every assistant, passed. Only a collection’s report gives this result. | How results are measured |
| You win (won) | A buyer question where 80% or more of its graded answers, across every assistant, passed. Only a collection’s report gives this result. | How results are measured |
Built on Potluck Docs, our open source template.