Skip to content
Agentic Craft

Agent Actions

Tool calls, subagent orchestration, parallel execution, plan cards, clarifying questions, decision flows, and approval gates for agent-driven approval and product workflows.

Tool Calls

Expandable, human-readable action indicators. Each tool call shows a plain-language label — never a function signature. Click to reveal key-value details.

1.2s
0.4s
3.4s
0.9s
PropertySpec
Tool title14px sans, font-weight 400 (never bold)
Inner detail text12px, key in muted, value in foreground
IconsAll monochrome — no colored status indicators
LabelsHuman-readable only — no function signatures or code
DurationRight-aligned in muted text; errors show "failed ·" prefix
InteractionClick anywhere on the header row to expand/collapse

Tool calls should feel like quiet status updates, not system logs. The user glances at them during execution and digs in only when something looks wrong.

Subagent Orchestration

Expandable rows showing delegated subagent work. Each row carries its own progress indicator and related tool calls.

Validating completeness and traceability of review source material for each review area.

x
1.8s
0.9s
4.2s
PropertySpec
Progress format"N of M unit" — concrete, not percentage-based
Completion"complete" badge replaces the progress count
Running indicatorPulsing dot next to the badge while in-progress
NestingNested tool calls inherit the same collapsed style
Progress bar1px foreground/40 bar with transition-transform duration-500

Subagent cards let users monitor delegated work without leaving their current context. The progress bar provides glanceable status while nested tools offer drill-down detail.

Plan Cards

Step-progress cards that show the agent's execution plan. Completed steps are muted with line-through; the current step pulses subtly; pending steps are dimmed.

Launch Coverage Report
3 of 6 complete
Load project brief
Parse requirement definitions
Map test cases to requirements
Identify coverage gaps
Generate report
Review with reviewer
PropertySpec
Step indicator6px square dot, rounded-sm — not a checkbox or number
Done stateMuted foreground + line-through
Active stateForeground + pulse animation
Pending stateMuted foreground at 60% opacity
Editable modeDrag handle + remove button on hover per step
Progress badge"N of M complete" or step count in editable mode

Plan cards give users a sense of progress and predictability. The editable mode allows corrections before the agent commits to an approach — a key trust-building pattern.

Parallel Execution

IDE-style tree connectors for operations running concurrently. The parent row collapses the branch; each child expands to show tool call details inline.

10:44 AM · 1s
10:44 AM · 1s
10:44 AM · 1s
10:44 AM · 1s
ElementSpec
Vertical spine1px line from parent icon to last child, masked by bg-background behind each icon
L-connectorsrounded-bl-lg border-l + border-b connecting spine to each child
Icon maskingsize-6 bg-background circle behind size-5 icon creates clean spine breaks
Hover stateRow label and icon transition to text-foreground; timestamp fades in on right
Last childbg-background mask covers spine below the final L-bend
Sequential fallbackFlat list of bordered rows, no tree connectors

The tree view uses Perplexity-style connectors — a vertical spine with L-shaped branch lines, icons that mask the spine for clean breaks, and hover-revealed timestamps. Each child row is lightweight: no borders, no cards, just text and an expand chevron.

Decision Flow

When the agent needs the user to choose between options that have different consequences. Each option shows what will happen if selected.

How should I handle the missing audit log entries for Timestamp handling?

PatternSpec
Pending stateAll options clickable with hover effect and consequence preview
Resolved stateSelected option highlighted, others hidden, confirmation shown
Selected cardPrimary border + subtle primary background tint
Consequence previewsBorder-left callout with arrow indicator inside each option
Auto-selectionNever auto-select — always wait for explicit user choice

Decisions are the highest-trust pattern in the system — they grant the user explicit control over how the agent proceeds. Never bypass them with auto-selection or hidden defaults.

Clarifying Questions

When the agent is blocked by missing intent, ask a focused question with answerable options, optional free text, and a clear skip path. This keeps clarification out of a vague back-and-forth chat loop.

Question 1 of 1

Which launch areas should I inspect before drafting findings?

PropertySpec
Question shapeOne focused question with 2-5 answerable options
SelectionSingle or multi-select rows with explicit selected state
Other inputInline textarea for constraints that do not fit the options
Skip pathVisible skip control when the agent can continue with defaults
SubmitContinue disabled until an option or free-text answer exists
Answered stateSelected answers collapse into readable chips before work resumes

Clarifying questions should reduce ambiguity, not create a survey. Prefer choices when the agent can name the likely answers. Use the free-text row only for constraints, missing context, or edge cases.

Approval Gate

Human-in-the-loop confirmation before the agent performs a consequential action. The user reviews the action details and explicitly approves or denies.

I'd like to send the launch summary submission email. Please review and approve.

Send launch summary email

Please find attached the Launch Review Summary for the customer portal v3.1 project brief. This submission covers all review checklist items for enterprise release with standard support coverage. The review was conducted using the current internal launch checklist.

Locked payload

Recipient

project-team@acme.internal

Subject

Launch Review — customer portal v3.1 (enterprise release)

Affected object

Project launch summary email

Cost / time

$0.09 estimated / under 1 min

Consequence External communication is sent and cannot be recalled.

Rollback Follow-up correction only

PropertySpec
Action summaryShield icon + plain-language description of what the agent wants to do
Detail panelMuted background container with key-value or diff preview
Diff formatGreen-ish (+) for additions, red-ish (−) for removals, muted tones
Approve buttonPrimary fill — should feel deliberate, not default
Deny buttonGhost/outline — lower visual weight than approve
Result statesSuccess (primary border), denied (destructive border), with reset button

Approval gates are the most critical trust pattern in the system. They ensure the agent never takes consequential actions without explicit human confirmation — essential for high-trust workflows where mistakes have regulatory consequences.