You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Update patterns and components for enhanced agent interaction features
- Added new interaction patterns: Intent Preview, Clarifying Questions, Autonomy Dial, and Escalation Pathway.
- Updated existing patterns to include replay functionality.
- Refined text and links in the Hero and Footer components for clarity.
- Removed unused code and improved styling in various components.
- Updated .gitignore to include design documentation.
// ── Before the agent acts ────────────────────────────────
10
23
{
11
-
id: 'streaming-text',
12
-
title: 'Streaming text reveal',
13
-
tagline: 'text appears as it thinks',
24
+
id: 'intent-preview',
25
+
title: 'Intent preview',
26
+
tagline: 'show the plan first',
14
27
description:
15
-
'Render the response token by token with a blinking caret so the interface feels alive while the model is still generating.',
16
-
Demo: StreamingTextDemo,
17
-
code: streamingTextCode,
18
-
specName: 'Return Moment',
19
-
specUrl: 'https://agentic-ux.com',
28
+
'Before doing anything, the agent restates the goal and lays out the plan it intends to follow — so the user can approve or adjust before work begins.',
29
+
Demo: IntentPreviewDemo,
30
+
code: intentPreviewCode,
31
+
specName: 'Intent Preview',
32
+
specUrl: `${SPEC}/intent-preview`,
33
+
replayable: true,
34
+
},
35
+
{
36
+
id: 'clarifying-questions',
37
+
title: 'Clarifying questions',
38
+
tagline: 'narrow scope before the plan',
39
+
description:
40
+
'When the request is ambiguous, the agent asks one focused follow-up with inline options instead of guessing — narrowing scope before it commits to a plan.',
41
+
Demo: ClarifyingQuestionsDemo,
42
+
code: clarifyingQuestionsCode,
43
+
specName: 'Clarifying Questions',
44
+
specUrl: `${SPEC}/clarifying-questions`,
45
+
replayable: true,
20
46
},
47
+
{
48
+
id: 'autonomy-dial',
49
+
title: 'Autonomy dial',
50
+
tagline: 'choose what the agent can do',
51
+
description:
52
+
'Let the user set how much the agent acts on its own — from manual to fully automatic — and reflect the chosen level back in plain language.',
53
+
Demo: AutonomyDialDemo,
54
+
code: autonomyDialCode,
55
+
specName: 'Autonomy Dial',
56
+
specUrl: `${SPEC}/autonomy-dial`,
57
+
},
58
+
59
+
// ── While the agent works ────────────────────────────────
'A slide-out panel that surfaces the context the agent is working from, so the user can inspect and trust what informs each response.',
102
+
Demo: ContextPanelDemo,
103
+
code: contextPanelCode,
104
+
specName: 'Explainable Rationale',
105
+
specUrl: `${SPEC}/explainable-rationale`,
106
+
},
107
+
108
+
// ── After the agent acts ─────────────────────────────────
109
+
{
110
+
id: 'action-confirm',
111
+
title: 'Action audit & undo',
112
+
tagline: 'review, approve, undo',
113
+
description:
114
+
'State the action and its details up front, require approval for consequential changes, and keep an undo within reach after it runs.',
115
+
Demo: ActionConfirmDemo,
116
+
code: actionConfirmCode,
117
+
specName: 'Action Audit & Undo',
118
+
specUrl: `${SPEC}/action-audit`,
119
+
},
120
+
{
121
+
id: 'escalation-pathway',
122
+
title: 'Escalation pathway',
123
+
tagline: 'hand off to a human',
124
+
description:
125
+
'When the agent hits a limit, it says so plainly and offers a path forward — hand off to a person or try another approach — instead of failing silently.',
126
+
Demo: EscalationPathwayDemo,
127
+
code: escalationPathwayCode,
128
+
specName: 'Escalation Pathway',
129
+
specUrl: `${SPEC}/escalation-pathway`,
130
+
replayable: true,
131
+
},
132
+
{
133
+
id: 'streaming-text',
134
+
title: 'Streaming text reveal',
135
+
tagline: 'text appears as it thinks',
136
+
description:
137
+
'Render the response token by token with a blinking caret so the interface feels alive while the model is still generating.',
0 commit comments