state.ts ×1

Frontier kind: Code frontier

unlabeled · c_03fb505d9e6a

3199 tests · 7070 LOC · 25 files · introduces 0 tests · 711 LOC · 4 files

Introduces — evidence that enters the hierarchy at this concept

Code
4 ranges711 lines · 4 files
Tests
0 tests

Contains — complete concept membership

All code (extent)
487 ranges7070 lines · 25 files · Browse complete extent
All tests (intent)
3199 testsBrowse complete intent

Neighbourhood graph

The orange circle is the focus. Violet and green circles are every ancestor and descendant, broader and narrower, at any distance; blue squares and pink diamonds are the introduced files and exact introduced tests of every visible concept, not only the focus's. Arrows point from broader to narrower concepts and bridge only concepts omitted from this view. Undirected links show source or test introduction. Concept and file size follows LOC; exact test nodes use test-count units.

Introduced files, introduced tests, and structurally relevant concept specialization

In the embedded map, ordinary wheel input scrolls the page; use the visible controls to zoom and drag to pan. Open the full-screen map for canvas navigation: wheel pans, Ctrl/Command plus wheel zooms, and arrow keys pan when this region is focused. On touch screens, open the full-screen map to pan or pinch. If JavaScript or WebGL is unavailable, use the native relationship evidence on this page.

Graph controls are ready.

Interactive rendering requires JavaScript and WebGL. Use the native relationship evidence on this page while the interactive map is unavailable.

Native relationship evidence

Every exact file and test below is linked only from the concept that introduces it.

Introduced tests

Every collected test enters the hierarchy at exactly one concept.

No tests are introduced at this concept. Its intent tests are introduced by other concepts.

Introduced code

Every collected source range enters the hierarchy at exactly one concept.

4 files ranked by introduced lines: 711 introduced LOC across 4 ranges. Expand a file to inspect source; the > gutter marks introduced lines.

src/vs/platform/agentHost/common/state/protocol/channels-changeset/state.ts 277 introduced LOC · 1 range

Open complete file

1 > /*--------------------------------------------------------------------------------------------- state.ts
2 > * Copyright (c) Microsoft Corporation. All rights reserved.
3 > * Licensed under the MIT License. See License.txt in the project root for license information.
4 > *--------------------------------------------------------------------------------------------*/
5 >
6 > // allow-any-unicode-comment-file
7 > // DO NOT EDIT -- auto-generated by scripts/sync-agent-host-protocol.ts
8 >
9 > import type { StringOrMarkdown, FileEdit, ErrorInfo } from '../common/state.js';
10 >
11 > // ─── Changesets ──────────────────────────────────────────────────────────────
12 >
13 > /**
14 > * Catalogue entry describing one changeset the server can produce for a
15 > * session.
16 > *
17 > * Catalogue entries are intentionally lightweight — just enough to render a
18 > * chip or list row without subscribing. Full per-changeset detail
19 > * ({@link ChangesetState}) lives on the subscribable URI obtained by
20 > * expanding {@link uriTemplate}.
21 > *
22 > * @category Changesets
23 > */
24 > export interface Changeset {
25 > /** Human-readable label, e.g. `"Uncommitted Changes"`. */
26 > label: string;
27 > /**
28 > * RFC 6570 URI template. Clients parse the variables directly out of the
29 > * template using the standard `{name}` syntax — they are not redeclared
30 > * here.
31 > *
32 > * Only the following template shapes are defined by this protocol; any
33 > * other variable name MUST be ignored by clients (there is no
34 > * protocol-defined way to obtain values for unknown variables):
35 > *
36 > * | Variables in template | Meaning |
37 > * | ------------------------------------------- | ------------------------------------------------------------------------------------ |
38 > * | _(none)_ | A static, session-wide changeset. The template is itself a subscribable URI. |
39 > * | `{turnId}` | Per-turn slice. Expand with a `Turn.id` from the session. |
40 > * | `{originalTurnId}` and `{modifiedTurnId}` | Diff between two turns. Both variables MUST be present. |
41 > *
42 > * Future protocol versions MAY add new well-known variables.
43 > */
44 > uriTemplate: string;
45 > /** Optional longer description. */
46 > description?: string;
47 > /**
48 > * Advisory hint describing what kind of changeset this is, so clients can
49 > * group, sort, or render an appropriate icon without parsing
50 > * {@link uriTemplate}. Recognized values include:
51 > *
52 > * - `'session'`: a static, session-wide changeset covering all changes the
53 > * agent has produced in this session.
54 > * - `'branch'`: changes relative to a base branch (e.g. a feature branch
55 > * diffed against `main`).
56 > * - `'uncommitted'`: the workspace's current uncommitted changes.
57 > * - `'turn'`: changes produced by a single turn. Typically paired with a
58 > * `{turnId}` variable in {@link uriTemplate}.
59 > * - `'compare-turns'`: a diff between two turns. Typically paired with
60 > * `{originalTurnId}` and `{modifiedTurnId}` variables in
61 > * {@link uriTemplate}.
62 > *
63 > * Implementations MAY provide additional values; clients SHOULD fall back
64 > * to a reasonable default when an unknown value is encountered.
65 > */
66 > changeKind: string;
67 > /**
68 > * Optional capability declarations for this changeset. Absent (or an empty
69 > * object) means the changeset advertises no optional capabilities.
70 > *
71 > * Because the catalogue entry is delivered up-front on
72 > * {@link ChangesetState | the session's changeset list}, clients can decide
73 > * whether to surface capability-gated UI (such as review checkboxes) without
74 > * first subscribing to the changeset URI. Mirrors the presence-flag
75 > * convention of `ClientCapabilities`.
76 > */
77 > capabilities?: ChangesetCapabilities;
78 > }
79 >
80 > /**
81 > * Optional capabilities a changeset advertises on its catalogue
82 > * {@link Changeset} entry.
83 > *
84 > * Each field is a presence flag: an empty object `{}` means "supported",
85 > * absence means "not supported". Sub-fields on individual capabilities are
86 > * reserved for future per-capability options.
87 > *
88 > * @category Changesets
89 > */
90 > export interface ChangesetCapabilities {
91 > /**
92 > * The changeset supports the per-file **review** workflow. When declared,
93 > * clients MAY surface a GitHub-style "Viewed" toggle per file and dispatch
94 > * {@link ChangesetFilesReviewChangedAction | `changeset/filesReviewChanged`} to
95 > * set each file's {@link ChangesetFile.reviewed} flag. Clients that omit
96 > * handling MUST treat the changeset as non-reviewable.
97 > */
98 > review?: Record<string, never>;
99 > }
100 >
101 > /**
102 > * Computation lifecycle of a {@link ChangesetState}.
103 > *
104 > * @category Changesets
105 > */
106 > export const enum ChangesetStatus {
107 > /** The server is still computing the contents of this changeset. */
108 > Computing = 'computing',
109 > /** The changeset has been fully computed and is up-to-date. */
110 > Ready = 'ready',
111 > /**
112 > * Computation failed. The cause is described by
113 > * {@link ChangesetState.error}.
114 > */
115 > Error = 'error',
116 > }
117 >
118 > /**
119 > * Full state for a single changeset, returned when a client subscribes to
120 > * an expanded changeset URI.
121 > *
122 > * The client already knows the URI it subscribed to, so this state does
123 > * not redundantly carry it (or the catalogue's `id`, `label`, etc.).
124 > * Aggregate counts (`additions`, `deletions`, `files`) are likewise
125 > * omitted: clients trivially compute them from `files[].edit.diff`.
126 > *
127 > * @category Changesets
128 > */
129 > export interface ChangesetState {
130 > /** Computation lifecycle. */
131 > status: ChangesetStatus;
132 > /** Present iff `status === ChangesetStatus.Error`. */
133 > error?: ErrorInfo;
134 > /** Files in this changeset, keyed by {@link ChangesetFile.id}. */
135 > files: ChangesetFile[];
136 > /**
137 > * Operations the client may invoke against this changeset. Omit when no
138 > * operations are available.
139 > */
140 > operations?: ChangesetOperation[];
141 > }
142 >
143 > /**
144 > * One file entry within a {@link ChangesetState}.
145 > *
146 > * @category Changesets
147 > */
148 > export interface ChangesetFile {
149 > /**
150 > * Stable identifier within the changeset. Typically `after.uri`
151 > * (or `before.uri` for deletions).
152 > */
153 > id: string;
154 > /**
155 > * Reuses the existing {@link FileEdit} shape. Clients derive line
156 > * additions, deletions, and rename/create/delete semantics from this.
157 > */
158 > edit: FileEdit;
159 > /**
160 > * Whether a reviewer has marked this file as reviewed (the GitHub-style
161 > * "Viewed" checkbox). Absent is equivalent to `false` — clients MUST treat
162 > * a missing value as not-yet-reviewed.
163 > *
164 > * Requires the changeset to advertise {@link ChangesetCapabilities.review}.
165 > * Clients toggle it by dispatching
166 > * {@link ChangesetFilesReviewChangedAction | `changeset/filesReviewChanged`};
167 > * the server MAY also originate it (e.g. an agent self-reviewing its own
168 > * output).
169 > *
170 > * There is no content version in the protocol, so review is **not** reset
171 > * automatically when a file's contents change under a stable id. The server,
172 > * which is the authority on what changed, resets review explicitly — either
173 > * by re-emitting the file (via {@link ChangesetFileSetAction} or
174 > * {@link ChangesetContentChangedAction}) without `reviewed: true`, or by
175 > * dispatching `changeset/filesReviewChanged` with `reviewed: false`.
176 > */
177 > reviewed?: boolean;
178 > /**
179 > * Server-defined opaque metadata, surfaced to operations and tooling
180 > * but not interpreted by the protocol.
181 > */
182 > _meta?: Record<string, unknown>;
183 > }
184 >
185 > /**
186 > * Execution lifecycle of a {@link ChangesetOperation}.
187 > *
188 > * An operation is invoked imperatively via `invokeChangesetOperation`, but
189 > * its progress and outcome are reflected back into changeset state so that
190 > * every subscriber observes a consistent view (e.g. a spinner on a "Create
191 > * Pull Request" button, or an inline error after a failed "revert").
192 > *
193 > * @category Changesets
194 > */
195 > export const enum ChangesetOperationStatus {
196 > /**
197 > * The operation is ready to be invoked. This is the default when
198 > * {@link ChangesetOperation.status} is omitted.
199 > */
200 > Idle = 'idle',
201 > /** An invocation of this operation is currently in flight. */
202 > Running = 'running',
203 > /**
204 > * The most recent invocation failed. The cause is described by
205 > * {@link ChangesetOperation.error}.
206 > */
207 > Error = 'error',
208 > /**
209 > * The operation is currently disabled and cannot be invoked.
210 > */
211 > Disabled = 'disabled',
212 > }
213 >
214 > /**
215 > * Where a {@link ChangesetOperation} can be invoked.
216 > *
217 > * @category Changesets
218 > */
219 > export const enum ChangesetOperationScope {
220 > /** Applies to the whole changeset. */
221 > Changeset = 'changeset',
222 > /** Applies to a single file within the changeset. */
223 > Resource = 'resource',
224 > /** Applies to a line range within a single file. */
225 > Range = 'range',
226 > }
227 >
228 > /**
229 > * A server-declared invokable verb the client can run against a
230 > * changeset, a file, or a range — `"stage"`, `"revert"`, `"create-pr"`,
231 > * and so on.
232 > *
233 > * The term "operation" is used deliberately to avoid colliding with the
234 > * protocol-level [Actions](/guide/actions) that mutate state.
235 > *
236 > * @category Changesets
237 > */
238 > export interface ChangesetOperation {
239 > /** Stable identifier, unique within this changeset. */
240 > id: string;
241 > /** Human-readable button/menu label. */
242 > label: string;
243 > /** Optional longer description shown on hover or in tooltips. */
244 > description?: string;
245 > /** Where this operation can be invoked. */
246 > scopes: ChangesetOperationScope[];
247 > /**
248 > * Optional confirmation prompt to show before invoking. When present,
249 > * the client MUST display this message to the user (typically in a
250 > * confirmation dialog) and only invoke the operation after the user
251 > * accepts. The presence of this field also signals that the operation
252 > * is destructive — clients SHOULD style the affirmative button
253 > * accordingly (e.g. with a warning colour).
254 > */
255 > confirmation?: StringOrMarkdown;
256 > /** Optional generic icon hint, e.g. `"check"`, `"trash"`. */
257 > icon?: string;
258 > /** Optional group identifier, used to group related operations together. */
259 > group?: string;
260 > /**
261 > * Current execution status. The server sets
262 > * {@link ChangesetOperationStatus.Running | Running} while an invocation
263 > * is in flight, {@link ChangesetOperationStatus.Error | Error} when the
264 > * most recent invocation failed, and
265 > * {@link ChangesetOperationStatus.Idle | Idle} otherwise.
266 > *
267 > * Clients SHOULD reflect this state in the UI — e.g. disabling the
268 > * control or showing a spinner while `Running`, and surfacing
269 > * {@link error} while `Error`.
270 > */
271 > status: ChangesetOperationStatus;
272 > /**
273 > * Cause of failure. Present iff
274 > * `status === ChangesetOperationStatus.Error`; otherwise omitted.
275 > */
276 > error?: ErrorInfo;
277 > }
src/vs/platform/agentHost/common/state/protocol/channels-root/state.ts 244 introduced LOC · 1 range

Open complete file

1 > /*--------------------------------------------------------------------------------------------- state.ts
2 > * Copyright (c) Microsoft Corporation. All rights reserved.
3 > * Licensed under the MIT License. See License.txt in the project root for license information.
4 > *--------------------------------------------------------------------------------------------*/
5 >
6 > // allow-any-unicode-comment-file
7 > // DO NOT EDIT -- auto-generated by scripts/sync-agent-host-protocol.ts
8 >
9 > import type { ConfigSchema, JsonPrimitive, ProtectedResourceMetadata } from '../common/state.js';
10 > import type { TerminalInfo } from '../channels-terminal/state.js';
11 > import type { Customization } from '../channels-session/state.js';
12 >
13 > // ─── Root State ──────────────────────────────────────────────────────────────
14 >
15 > /**
16 > * Policy configuration state for a model.
17 > *
18 > * @category Root State
19 > */
20 > export const enum PolicyState {
21 > Enabled = 'enabled',
22 > Disabled = 'disabled',
23 > Unconfigured = 'unconfigured',
24 > }
25 >
26 > /**
27 > * Global state shared with every client subscribed to `ahp-root://`.
28 > *
29 > * @category Root State
30 > */
31 > export interface RootState {
32 > /** Available agent backends and their models */
33 > agents: AgentInfo[];
34 > /** Number of active (non-disposed) sessions on the server */
35 > activeSessions?: number;
36 > /** Known terminals on the server. Subscribe to individual terminal URIs for full state. */
37 > terminals?: TerminalInfo[];
38 > /** Agent host configuration schema and current values */
39 > config?: RootConfigState;
40 > /**
41 > * Additional implementation-defined metadata about the agent host itself.
42 > *
43 > * Clients MAY look for well-known keys here to provide enhanced UI.
44 > */
45 > _meta?: Record<string, unknown>;
46 > }
47 >
48 > /**
49 > * @category Root State
50 > */
51 > export interface AgentInfo {
52 > /** Agent provider ID (e.g. `'copilot'`) */
53 > provider: string;
54 > /** Human-readable name */
55 > displayName: string;
56 > /** Description string */
57 > description: string;
58 > /** Available models for this agent */
59 > models: SessionModelInfo[];
60 > /**
61 > * Protected resources this agent requires authentication for.
62 > *
63 > * Each entry describes an OAuth 2.0 protected resource using
64 > * [RFC 9728](https://datatracker.ietf.org/doc/html/rfc9728) semantics.
65 > * Clients should obtain tokens from the declared `authorization_servers`
66 > * and push them via the `authenticate` command before creating sessions
67 > * with this agent.
68 > *
69 > * @see {@link /specification/authentication | Authentication}
70 > */
71 > protectedResources?: ProtectedResourceMetadata[];
72 > /**
73 > * Customizations associated with this agent.
74 > *
75 > * Either container customizations —
76 > * {@link PluginCustomization | `PluginCustomization`} entries the agent
77 > * bundles, plus {@link DirectoryCustomization | `DirectoryCustomization`}
78 > * entries it watches in any workspace it's used with — or top-level
79 > * {@link McpServerCustomization | `McpServerCustomization`} entries
80 > * the agent host declares directly. When a session is created with
81 > * this agent, these entries are augmented (e.g. directory URIs are
82 > * resolved against the workspace, children are parsed) and propagated
83 > * into the session's `customizations` list.
84 > */
85 > customizations?: Customization[];
86 > /**
87 > * Static capabilities the agent advertises about itself. Clients use these
88 > * to gate features (multi-chat, fork) instead of switching on the provider
89 > * id.
90 > */
91 > capabilities?: AgentCapabilities;
92 > }
93 >
94 > /**
95 > * Static capabilities an {@link AgentInfo} advertises. Modelled after MCP
96 > * capabilities: each field is opt-in and its presence (an empty object `{}`)
97 > * signals support, while absence means the feature is unsupported and the
98 > * corresponding client commands MUST NOT be used. Sub-fields carry
99 > * per-capability options.
100 > *
101 > * @category Root State
102 > */
103 > export interface AgentCapabilities {
104 > /**
105 > * The agent can host more than one concurrent chat per session. When absent,
106 > * clients MUST NOT call `createChat` to open chats beyond the default one the
107 > * session starts with. An empty object `{}` advertises multi-chat without
108 > * source-based creation; set {@link MultipleChatsCapability.fork} or
109 > * {@link MultipleChatsCapability.sideChat} to allow the corresponding mode.
110 > */
111 > multipleChats?: MultipleChatsCapability;
112 > /**
113 > * The session's agent can be granted tool access to more than one working
114 > * directory. The directories are treated as equal peers except where the
115 > * agent advertises {@link MultipleWorkingDirectoriesCapability.requiresPrimary}
116 > * (some backends need one directory designated as a primary root).
117 > *
118 > * When absent, clients MUST NOT mutate a session's or chat's working-directory
119 > * set and MUST NOT set more than one entry in
120 > * {@link CreateSessionParams.workingDirectories}.
121 > */
122 > multipleWorkingDirectories?: MultipleWorkingDirectoriesCapability;
123 > }
124 >
125 > /**
126 > * Options for the {@link AgentCapabilities.multipleChats} capability.
127 > *
128 > * @category Root State
129 > */
130 > export interface MultipleChatsCapability {
131 > /**
132 > * The agent can fork a chat from a specific turn. When absent or `false`,
133 > * clients MUST NOT pass a {@link ChatSource} with `kind: "fork"` to
134 > * `createChat`.
135 > * Forking always implies multi-chat support.
136 > */
137 > fork?: boolean;
138 > /**
139 > * The agent can create a side chat from a specific turn. When absent or
140 > * `false`, clients MUST NOT pass a {@link ChatSource} with
141 > * `kind: "sideChat"` to `createChat`.
142 > *
143 > * A side chat receives the source turn as context without copying the source
144 > * transcript into its own visible history. The source is identified by a
145 > * stable `turnId`, which the host resolves against the source chat's current
146 > * `activeTurn` or retained history. When it names the current active turn,
147 > * the host snapshots the available partial assistant response at creation
148 > * time. Side-chat support always implies multi-chat support.
149 > */
150 > sideChat?: boolean;
151 > }
152 >
153 > /**
154 > * Options for the {@link AgentCapabilities.multipleWorkingDirectories} capability.
155 > *
156 > * @category Root State
157 > */
158 > export interface MultipleWorkingDirectoriesCapability {
159 > /**
160 > * The agent requires each chat to designate one of its working directories as
161 > * the **primary** — a distinguished root the chat is centered on (e.g. the
162 > * agent's process root for that chat, the default location for relative
163 > * paths). Primary is a **per-chat** notion, fixed at chat creation. When
164 > * `true`, a client SHOULD supply {@link CreateChatParams.primaryWorkingDirectory}
165 > * (and {@link CreateSessionParams.primaryWorkingDirectory}, which seeds the
166 > * session's default chat); a host MAY reject creation that omits it, or fall
167 > * back to the first entry of the chat's working directories. The chosen
168 > * primary is reported (read-only) on {@link ChatState.primaryWorkingDirectory}.
169 > *
170 > * When absent or `false`, the agent has no primary — all directories are
171 > * equal peers and clients need not designate one.
172 > */
173 > requiresPrimary?: boolean;
174 > }
175 >
176 > /**
177 > * @category Root State
178 > */
179 > export interface SessionModelInfo {
180 > /** Model identifier */
181 > id: string;
182 > /** Provider this model belongs to */
183 > provider: string;
184 > /** Human-readable model name */
185 > name: string;
186 > /** Maximum context window size */
187 > maxContextWindow?: number;
188 > /** Maximum number of output tokens the model can generate */
189 > maxOutputTokens?: number;
190 > /** Maximum number of prompt (input) tokens the model accepts */
191 > maxPromptTokens?: number;
192 > /** Whether the model supports vision */
193 > supportsVision?: boolean;
194 > /** Policy configuration state */
195 > policyState?: PolicyState;
196 > /**
197 > * Configuration schema describing model-specific options (e.g. thinking
198 > * level). Clients present this as a form and pass the resolved values in
199 > * {@link ModelSelection.config} when creating or changing sessions.
200 > */
201 > configSchema?: ConfigSchema;
202 > /**
203 > * Additional provider-specific metadata for this model.
204 > *
205 > * Clients MAY look for well-known keys here to provide enhanced UI.
206 > * For example, a `pricing` key may carry model pricing metadata.
207 > */
208 > _meta?: Record<string, unknown>;
209 > }
210 >
211 > /**
212 > * A model selection: the chosen model ID together with any model-specific
213 > * configuration values whose keys correspond to the model's
214 > * {@link SessionModelInfo.configSchema}.
215 > *
216 > * @category Root State
217 > */
218 > export interface ModelSelection {
219 > /** Model identifier */
220 > id: string;
221 > /**
222 > * Model-specific configuration values. Values are JSON primitives: most
223 > * pickers produce strings, but some (e.g. a numeric context-size picker)
224 > * produce numbers or booleans, which are carried through as-is.
225 > */
226 > config?: Record<string, JsonPrimitive>;
227 > }
228 >
229 > // ─── Root Config Types ───────────────────────────────────────────────────────
230 >
231 > /**
232 > * Live agent-host configuration metadata.
233 > *
234 > * The schema describes the available configuration properties and the values
235 > * contain the current value for each resolved property.
236 > *
237 > * @category Root State
238 > */
239 > export interface RootConfigState {
240 > /** JSON Schema describing available configuration properties */
241 > schema: ConfigSchema;
242 > /** Current configuration values */
243 > values: Record<string, unknown>;
244 > }
src/vs/platform/agentHost/common/state/protocol/channels-terminal/state.ts 173 introduced LOC · 1 range

Open complete file

1 > /*--------------------------------------------------------------------------------------------- state.ts
2 > * Copyright (c) Microsoft Corporation. All rights reserved.
3 > * Licensed under the MIT License. See License.txt in the project root for license information.
4 > *--------------------------------------------------------------------------------------------*/
5 >
6 > // allow-any-unicode-comment-file
7 > // DO NOT EDIT -- auto-generated by scripts/sync-agent-host-protocol.ts
8 >
9 > import type { URI } from '../common/state.js';
10 >
11 > // ─── Terminal Types ──────────────────────────────────────────────────────────
12 >
13 > /**
14 > * Lightweight terminal metadata exposed on the root state.
15 > *
16 > * @category Terminal Types
17 > */
18 > export interface TerminalInfo {
19 > /** Terminal URI (subscribable for full terminal state) */
20 > resource: URI;
21 > /** Human-readable terminal title */
22 > title: string;
23 > /** Who currently holds this terminal */
24 > claim: TerminalClaim;
25 > /** Process exit code, if the terminal process has exited */
26 > exitCode?: number;
27 > }
28 >
29 > /**
30 > * Discriminant for terminal claim kinds.
31 > *
32 > * @category Terminal Types
33 > */
34 > export const enum TerminalClaimKind {
35 > Client = 'client',
36 > Session = 'session',
37 > }
38 >
39 > /**
40 > * A terminal claimed by a connected client.
41 > *
42 > * @category Terminal Types
43 > */
44 > export interface TerminalClientClaim {
45 > /** Discriminant */
46 > kind: TerminalClaimKind.Client;
47 > /** The `clientId` of the claiming client */
48 > clientId: string;
49 > }
50 >
51 > /**
52 > * A terminal claimed by a session, optionally scoped to a specific turn or tool call.
53 > *
54 > * @category Terminal Types
55 > */
56 > export interface TerminalSessionClaim {
57 > /** Discriminant */
58 > kind: TerminalClaimKind.Session;
59 > /** Session URI that claimed the terminal */
60 > session: URI;
61 > /** Optional turn identifier within the session */
62 > turnId?: string;
63 > /** Optional tool call identifier within the turn */
64 > toolCallId?: string;
65 > }
66 >
67 > /**
68 > * Describes who currently holds a terminal. A terminal may be claimed by
69 > * either a connected client or a session (e.g. during a tool call).
70 > *
71 > * @category Terminal Types
72 > */
73 > export type TerminalClaim = TerminalClientClaim | TerminalSessionClaim;
74 >
75 > /**
76 > * Full state for a single terminal, loaded when a client subscribes to the terminal's URI.
77 > *
78 > * @category Terminal Types
79 > */
80 > export interface TerminalState {
81 > /** Human-readable terminal title */
82 > title: string;
83 > /** Current working directory of the terminal process */
84 > cwd?: URI;
85 > /** Terminal width in columns */
86 > cols?: number;
87 > /** Terminal height in rows */
88 > rows?: number;
89 > /**
90 > * Typed content parts, replacing the flat `content: string`.
91 > *
92 > * Naive consumers that only need the raw VT stream can reconstruct it with:
93 > * `content.map(p => p.type === 'command' ? p.output : p.value).join('')`
94 > *
95 > * Consumers that need command boundaries can filter by part type.
96 > */
97 > content: TerminalContentPart[];
98 > /** Process exit code, set when the terminal process exits */
99 > exitCode?: number;
100 > /** Who currently holds this terminal */
101 > claim: TerminalClaim;
102 > /**
103 > * Whether this terminal emits `terminal/commandExecuted` and
104 > * `terminal/commandFinished` actions and populates `command`-typed parts.
105 > *
106 > * Clients MUST check this flag before relying on command detection.
107 > * Do NOT use the presence of a `command` part as a feature flag — parts
108 > * are absent in the normal idle state.
109 > */
110 > supportsCommandDetection?: boolean;
111 > /**
112 > * Whether this terminal-style resource is backed by a pseudoterminal.
113 > * When `false`, output is plain text and clients do not need to parse
114 > * VT sequences.
115 > */
116 > isPty?: boolean;
117 > }
118 >
119 > // ─── Terminal Content Parts ──────────────────────────────────────────────────
120 >
121 > /**
122 > * A content part within terminal output.
123 > *
124 > * @category Terminal Types
125 > */
126 > export type TerminalContentPart =
127 > | TerminalUnclassifiedPart
128 > | TerminalCommandPart;
129 >
130 > /**
131 > * Unstructured terminal output — content before, between, or after commands,
132 > * or from terminals that do not support command detection.
133 > *
134 > * @category Terminal Types
135 > */
136 > export interface TerminalUnclassifiedPart {
137 > type: 'unclassified';
138 > /** Accumulated VT output. Appended to by `terminal/data` when no command is executing. */
139 > value: string;
140 > }
141 >
142 > /**
143 > * A single command: its command line and the output it produced.
144 > *
145 > * While `isComplete` is false the command is still executing; `output` grows
146 > * as `terminal/data` actions arrive. At `terminal/commandFinished` the part
147 > * is mutated in-place with `isComplete: true` and the completion metadata.
148 > *
149 > * @category Terminal Types
150 > */
151 > export interface TerminalCommandPart {
152 > type: 'command';
153 > /**
154 > * Stable id matching the `commandId` on the corresponding
155 > * `terminal/commandExecuted` and `terminal/commandFinished` actions.
156 > */
157 > commandId: string;
158 > /** The command line submitted to the shell. */
159 > commandLine: string;
160 > /**
161 > * Accumulated VT output. Appended to by `terminal/data` while `isComplete`
162 > * is false. Shell integration escape sequences are stripped by the server.
163 > */
164 > output: string;
165 > /** Unix timestamp (ms) when execution started, as reported by the server. */
166 > timestamp: number;
167 > /** Whether the command has finished. */
168 > isComplete: boolean;
169 > /** Shell exit code. Set at completion. `undefined` if unknown. */
170 > exitCode?: number;
171 > /** Wall-clock duration in milliseconds. Set at completion. */
172 > durationMs?: number;
173 > }
src/vs/platform/agentHost/common/state/protocol/state.ts 17 introduced LOC · 1 range

Open complete file

1 > /*--------------------------------------------------------------------------------------------- state.ts
2 > * Copyright (c) Microsoft Corporation. All rights reserved.
3 > * Licensed under the MIT License. See License.txt in the project root for license information.
4 > *--------------------------------------------------------------------------------------------*/
5 >
6 > // allow-any-unicode-comment-file
7 > // DO NOT EDIT -- auto-generated by scripts/sync-agent-host-protocol.ts
8 >
9 > export * from './common/state.js';
10 > export * from './channels-root/state.js';
11 > export * from './channels-session/state.js';
12 > export * from './channels-chat/state.js';
13 > export * from './channels-terminal/state.js';
14 > export * from './channels-changeset/state.js';
15 > export * from './channels-annotations/state.js';
16 > export * from './channels-otlp/state.js';
17 > export * from './channels-resource-watch/state.js';