| Chat | General conversation, model replies, and Tool entry |
| Coding | Local Workspace, files, commands, Git, and validation |
| Team | Multi-Agent collaboration surface |
| Workspace | Project scope for reading, changing, and validating files |
| Provider | Service or connection that handles model requests |
| Model | Concrete model exposed by a Provider |
| Tool | Capability that performs a read, write, network, browser, or other action |
| Skill | Reusable method, constraints, templates, scripts, and requirements |
| MCP Server | Service exposing MCP Tools, Resources, or Templates |
| Resource | Readable external content from an MCP Server |
| Descriptor | Tool name, description, arguments, Schema, risk, and scope |
| Quarantine | Temporarily isolated capability after a descriptor or Schema change |
| Plugin | Declarative package for Skills, Commands, MCP, and Hooks |
| Command | Markdown/TOML workflow entry |
| Slash Command | Command invoked with a slash |
| Hook | Logic that runs, changes, or blocks a lifecycle event |
| Memory | Information retrievable by later tasks |
| Context | Messages, Tools, Skills, files, Memory, and external results for a request |
| Compaction | Summary of earlier context to free space |
| Token | Unit used to measure model input, output, and Tool descriptions |
| Tool approval | Approval for one concrete Tool action |
| Effect class | Read-only, idempotent, non-idempotent, or unknown side-effect class |
| Agent Profile | Identity, capabilities, settings, and scope of an Agent |
| Work Item | Scoped Team unit with state and deliverable |
| Artifact | File, result, or change produced by a task |
| Live Memory | Current Team context |
| Archived Memory | Historical context not automatically treated as current |
| Activity Recorder | Local screen text, OCR, summary, and optional visual history |
| Computer Use | Permissioned operation of local graphical applications |