Glossary
Terms used across this documentation, with the page that describes each. Where the product shows a plainer word than the underlying model (transformation for strategy, data point for storage, Requirements for features), both are listed.
| Term | Meaning | Page |
|---|---|---|
| project | One governed unit of work: a configuration database, a paired engine fork, its apps, agents, workflows, docs and tests, scoped by a slug. | Platform overview |
| fork | The project's own copy of the engine, prototyped from the core; holds its endpoints, services, app services and vault. | Architecture and components |
| core (prototype) | The shared engine project every fork inherits from. | Architecture and components |
| endpoint | A named flow graph on the engine; what a tool, a test or an app calls. | Architecture and components |
| service | A container the engine manages for a fork. | Architecture and components |
| function service | A service where the platform generates a small server around handler code. | Architecture and components |
| app service | An arbitrary application with its own image definition and source tree, built and run by the service manager; the default shape of every new app. | Independent applications |
| view | The older app shape: a screen defined by an endpoint and rendered inside the cortex UI. | Applications |
| studio | The embedded engine console for a fork's endpoints, services and app services. | Architecture and components |
| node | The authoring unit of a pipeline: a load, save or transform node. | Data governance and pipelines |
| link | The connection between two nodes on the pipeline canvas. | Data governance and pipelines |
| storage (data point) | Any named place data lives, with a schema: a source, a target or an intermediate. | Data governance and pipelines |
| strategy (transformation) | One typed step from a domain storage to a codomain storage, backed by one endpoint per kind. | Data governance and pipelines |
| data element | A governed field definition source columns map to and the target binds. | Data governance and pipelines |
| processing rule | A rule in the project's configuration applied inline while records stream through the pipeline. | Data governance and pipelines |
| reference list | A governed list of allowed values a rule or mapping refers to. | Data governance and pipelines |
| data source | A connection or file the pipeline reads: an upload, a database, or a master-built source. | Data governance and pipelines |
| data target | A destination the pipeline writes, typically a relational database. | Data governance and pipelines |
| pipeline run | One execution of the transformation graph built from the nodes and links. | Data governance and pipelines |
| recipe | The whole pipeline configuration of a project as one versioned set. | Data governance and pipelines |
| configuration store | The versioned database with commit history and branches that holds each recipe. | Architecture and components |
| branch | A branch of a project's recipe, selected per request by the branch header. | Integrations and API |
| agent | A manifest on the fork whose tools are fork endpoints, run per chat turn or delegated task. | Agents and MCP |
| manifest | The endpoint that defines an agent's identity, persona, model, tools and safeguards. | Agents and MCP |
| persona | One of the three built-in harnesses: governance, engine, analyst. | Agents and MCP |
| tool | A typed operation an agent can call, exposed over MCP. | Agents and MCP |
| safeguards | The manifest block that bounds an agent: scope, grounding, verification, tool budget. | Safeguards and correctness |
| unsupported claim | A statement in an agent's draft that no tool result backs. | Safeguards and correctness |
| ledger | The record of guarded answers, or of a requirement's tasks. | Requirements and knowledge |
| orchestrator | The chat persona that delegates to visible agents through one delegation tool per agent. | Agents and MCP |
| delegate | An agent the orchestrator can hand a task to in the current chat. | Agents and MCP |
| pinned chat | A chat whose roster is one agent. | Agents and MCP |
| actor | An agent that executes workflow steps; its manifest carries an actor block. | Business workflows |
| execution contract | The part of an actor's persona that says what to produce per step. | Business workflows |
| workflow | A business process drawn as steps, decisions and ends assigned to roles, users or actors. | Business workflows |
| definition | The saved shape of a workflow on the canvas. | Business workflows |
| run | One live execution of a workflow definition. | Business workflows |
| listener | A workflow trigger: an email listener, the upload form, a schedule, a data condition. | Business workflows |
| My Work | The inbox of workflow steps assigned to a person. | Business workflows |
| fleet | The pool of coding-agent worker sessions that claim Board cards and generation jobs. | Delivery lifecycle |
| card | One request on the Board: a new app, a change, a bug fix, a build from sources. | Delivery lifecycle |
| Board | The nav item where requests are filed, claimed and completed. | Delivery lifecycle |
| requirement source (material) | An uploaded file or typed request in Docs, versioned as a set. | Requirements and knowledge |
| requirement (feature) | One row of what the project must do or does, with kind, origin, use cases and status. | Requirements and knowledge |
| kind | A requirement's class: functional, quality, constraint, design, non-goal or noise. | Requirements and knowledge |
| origin | Where a requirement came from: a passage of a source, or the live application (as-built). | Requirements and knowledge |
| coverage | The rule that every passage of every source is linked to a row. | Requirements and knowledge |
| preview | The build preview: variants with screen mockups approved before the build card is filed. | Requirements and knowledge |
| task | Work created from a requirement and filed to the Board. | Requirements and knowledge |
| evidence | Screenshots, commits, test notes, steps or links on an implemented task. | Requirements and knowledge |
| test | A declarative specification linked to a requirement, run by the portable runner. | Tests and health |
| health | The one verdict per project (green, red, none) folded from the newest test results. | Tests and health |
| watchdog | The schedule that runs every project's tests every fifteen minutes, after boot and after a deploy. | Tests and health |
| Status page | The Operations page showing every test's newest verdict and the run history. | Tests and health |
| companion | A cortex feature an app carries in its speed dial: Chat, Workflows, Guides, Models, Pipeline, Governance, Docs. | Applications |
| viewer options | The owner's choice of companions and chat kind for an app. | Applications |
| share link | The link that admits an outside person to one app as an application user. | Applications |
| application user | A person admitted by a share alone: the app and its companions, no project access. | Security and operations |
| app SDK | The script an app service includes with one tag to render its speed dial and sign-in gate. | Integrations and API |
| guide | A saved spotlight tour over an app, with an optional PDF. | Requirements and knowledge |
| knowledge page | A Markdown entry in the fork vault, linked by wikilinks, shown under Docs, Knowledge. | Requirements and knowledge |
| vault | The fork's store of Markdown entries and their link graph. | Requirements and knowledge |
| summary | The generated customer-facing account of a project, with editable notes. | Requirements and knowledge |
| model | A trained prediction (classification, regression or time series) with versions and an algorithm. | Models |
| deployment | A project shipped as a standalone application in its own namespace, or a downloadable archive. | Deployments and environments |
| delivery | The per-project options on the Delivery card that decide what an export ships. | Deployments and environments |
| customer archive | The self-contained bundle a customer deploys themselves, with per-archive secrets and a runbook. | Deployments and environments |
| distribute | The delivery flags, one per companion and all off by default. | Deployments and environments |
| development stand-in | A sample dependency shipped for development only, never in production files. | Deployments and environments |
| runbook | The operations document generated from the service definitions and shipped with every delivery. | Deployments and environments |
| environment | One installation of cortex, such as development or production. | Deployments and environments |
| access token | A labelled MCP bearer minted in Settings for a headless client. | Integrations and API |
| MCP | The Model Context Protocol; cortex serves its tools over Streamable HTTP behind OAuth 2.1. | Agents and MCP |
| Feed | The page where meeting transcripts and Slack digests are collected and routed into a project. | Integrations and API |