AI-Assisted Project Editor¶

The project editor includes AI assistance that helps you generate, refine, and validate YAML configurations using natural language. Instead of memorizing YAML schemas, describe what you want and let AI handle the syntax.

Three Modes¶

Generate (start from a description)¶

When you create a new dashboard, query, agent, or any configuration, the editor shows a prompt field instead of blank YAML:

Click + New dashboard (or any editor type) in the tree sidebar
Type what you need: "Show team velocity by sprint with a burndown chart"
Click Generate --- AI creates complete, valid YAML
Review the output and click Apply to load it into the editor
Fine-tune the YAML or use Refine mode for further changes

AI uses your project context (workflow states, custom fields, installed extensions, existing queries) to generate configurations that reference real data in your project.

Refine (modify existing YAML)¶

For any existing configuration, click the "Ask AI to refine..." bar above the YAML editor:

Click the prompt trigger (sparkle icon) above the editor
Type your change: "Add a two-person approval gate before the deploy state"
AI generates a diff showing exactly what changes
Review the suggestion with confidence score and description
Click Apply to merge the changes, or refine further

Refine mode preserves your existing YAML and makes minimal, targeted changes. You always see what changed before applying.

Validate (catch issues as you edit)¶

The editor continuously validates your YAML in the background:

Syntax errors (tabs, invalid YAML) are caught instantly
Logical issues (orphan states, circular transitions, unreachable nodes) are detected by AI
Contextual warnings (WIP limits vs. team size, missing extension references) use project context
Industry compliance (healthcare workflow missing audit state, SOX workflow missing approval gate)

Each issue shows a severity level and an optional Fix button that applies the corrected YAML with one click.

Supported Editor Types¶

AI assistance works for all 7 editor types:

Editor	Generate Example	Refine Example
Workflow	"Create an employee onboarding workflow"	"Add SLA timers to the review state"
Board	"Create a sprint board with code review column"	"Add swimlanes grouped by assignee"
Intake Form	"Create a bug report form with severity dropdown"	"Add file upload and environment fields"
Automation	"Auto-assign critical bugs to the on-call engineer"	"Add Slack notification when SLA breaches"
Dashboard	"Show velocity, cycle time, and blocker count"	"Add a pie chart for issue distribution by type"
Query	"Average resolution time grouped by priority"	"Filter to last 90 days and add trend line"
Agent	"Create a sprint planning assistant"	"Add team capacity to the agent's context"

Cross-Editor Awareness¶

AI knows about your entire project configuration. When generating a dashboard, it references existing queries. When adding workflow actions, it suggests installed extensions. When configuring an agent, it considers your team size and custom fields.

This means suggestions are specific to your project, not generic templates.

AI Chat Panel¶

The project editor includes a collapsible AI panel on the right side with:

Proactive Suggestions¶

Cards that appear automatically based on your current configuration: - Schema improvements --- missing WIP limits, orphan states, undefined references - Extension integration --- installed extensions not referenced in this config - Marketplace templates --- relevant templates matching your editor type and industry - Similar repositories --- public GitHub repos with similar .kiket/ configurations

Each card has an action button (Apply, Install, View) --- no suggestion is informational-only.

Quick Actions¶

Chips for common operations per editor type: - Workflow: Add state, Add approval gate, Add SLA timer, Add checklist - Board: Add column, Set WIP limits, Add swimlanes - Dashboard: Add chart, Add KPI, Set refresh interval

Click a chip to send it as a chat message.

Multi-Turn Chat¶

A persistent chat at the bottom of the panel for conversational editing: - Maintains conversation history across messages - YAML code blocks with Apply button to update the editor - Follow-up suggestion chips after each response - Context-aware --- the AI knows your project, installed extensions, and current YAML

Keyboard Shortcuts¶

Key	Action
`Cmd+Shift+I`	Toggle AI panel

Keyboard Shortcuts¶

Key	Action
`Enter`	Submit prompt
`Shift+Enter`	New line in prompt
`Escape`	Close prompt bar

Privacy¶

AI processes your YAML configuration and project schema (field names, state names, extension names)
Actual issue data, user data, and file contents are not sent to the AI
All AI operations are logged in the audit trail
See AI Overview for data privacy details