Skip to main content
Implementations transform your Plans (PRDs) into concrete, actionable work items. Using AI and semantic code search, Kasava generates structured issues organized into epics that can be exported to GitHub, Linear, Jira, or Asana.

What is an Implementation?

An Implementation is an actionable breakdown of a Plan containing:
  • Epics - Logical groupings of related work (e.g., “Backend API”, “Frontend UI”, “Testing”)
  • Issues - Individual work items with descriptions, acceptance criteria, and code context
  • Code Context - AI-identified files and patterns relevant to each issue
  • Export Capability - Bulk export to external project management platforms

Product Planning Hierarchy

Implementations fit into Kasava’s planning hierarchy:
Product
├── Initiatives (strategic themes)
│     └── Plans (requirements documents)
│           └── Implementations (work breakdown)
│                 └── Epics (logical groupings)
│                       └── Issues (individual tasks)

└── Plans (standalone)
      └── Implementations

Accessing Implementations

You can access Implementations in two ways:

From the Implementations Page

1

Open the Sidebar

Click Implementations in the main sidebar navigation to go to the Implementations list
2

Browse Your Implementations

View all your implementations displayed as cards in a responsive grid layout
3

Click to Open

Click any implementation card to view its details
Implementations list page showing cards with status badges The list view displays:
  • Implementation cards - Each showing title, description, status, epic count, and issue count
  • Status badges - Draft, In Progress, Completed, or Archived
  • Source Plan - Link to the originating Plan
  • Creation date - When the Implementation was created

From a Plan

When viewing a Plan, you can quickly access its implementation or create a new one.

Creating an Implementation

Implementations are generated from Plans using AI:
1

Open Your Plan

Navigate to the Plan you want to implement from the Plans page in the sidebar
2

Click Generate Issues

In the Plan header, click the Generate Issues button. If the Plan already has an implementation, you’ll see a View Issues button instead that takes you directly to the existing implementation.
3

Select Repository (Optional)

Choose a repository from your organization to enable code context analysis. This helps the AI understand your codebase structure and generate more specific, relevant issues.
4

Configure Options

Set generation options:
  • Include code context - Analyze repository code to generate more relevant issues (requires repository selection)
  • Include tests - Generate test-related issues for the implementation
5

Generate

Click Generate to start the AI analysis process
Implementation Generation dialog with repository selection and options

Generation Progress

During generation, you’ll see progress through these stages:
  1. Analyzing Plan requirements - Parsing your Plan content
  2. Loading repository context - Fetching repository structure
  3. Searching codebase for relevant files - Semantic code search
  4. Generating issues with AI - Creating structured issues
  5. Saving implementation - Persisting the results
Generation progress dialog showing stages

What the AI Does

During generation, the AI:
  1. Parses Plan Content - Extracts requirements, user stories, and success criteria from your Plan
  2. Analyzes Repository - Identifies tech stack, patterns, and directory structure
  3. Performs Code Search - Finds relevant files for each requirement using semantic search
  4. Generates Issues - Creates structured issues with code context and implementation hints
  5. Organizes into Epics - Groups related issues logically by component or feature area

Viewing an Implementation

Click on any Implementation card to view its details. Implementation detail page showing epics and issues The detail view shows:

Header Information

  • Title and description - Implementation overview
  • Status badge - Current state (Draft, In Progress, Completed, Archived)
  • Source Plan link - Click to navigate back to the originating Plan
  • Creation date - When the Implementation was generated
  • Back button - Return to the Implementations list

Statistics Panel

A summary panel shows key metrics at a glance:
MetricDescription
EpicsTotal number of epic groupings
Total IssuesAll issues in the Implementation
SelectedIssues currently selected for export
EstimatedTotal estimated hours for all issues

Epic Sections

Each epic is displayed as a collapsible section showing:
  • Epic name and color - Visual identifier
  • Selection count - How many issues are selected (e.g., “5/8 selected”)
  • Issue list - All issues within the epic

Reviewing Issues

After generation, review and edit issues before export:

Issue Details

Each generated issue includes:
FieldDescription
TitleAction-oriented task description
DescriptionImplementation details (collapsible)
PriorityLow, Medium, High, or Critical
Estimated HoursTime estimate for completion
StatusDraft or Exported
Expanded issue showing all fields

Issue Row

Each issue row displays:
  • Checkbox - Toggle selection for export
  • Title - Task name (truncated if long)
  • Priority badge - Color-coded priority indicator (Low, Medium, High, or Critical)
  • Exported badge - Shows “Exported” if the issue has already been sent to a platform
  • Description - Truncated preview of the issue description
  • Estimated hours - Time estimate displayed on the right (e.g., “4h”)

Managing Selection

Control which issues get exported by using the checkboxes:
  • Click the checkbox next to any issue to toggle its selection
  • The epic header shows the current selection count (e.g., “3/5 selected”)
  • Only selected issues will be included when you export to an external platform

Epics

Epics group related issues for better organization:

Default Epics

AI typically generates epics based on your codebase structure:
  • Data Model - Database schema and migrations
  • Backend API - API endpoints and services
  • Frontend UI - User interface components
  • Business Logic - Core application logic
  • Testing - Unit, integration, and e2e tests
  • Documentation - Technical and user documentation

Epic Display

Each epic section shows:
  • Collapsible header - Click to expand/collapse
  • Color indicator - Visual categorization
  • Epic name - Descriptive title
  • Selection badge - Count of selected issues
  • Description - Epic overview (if provided)
  • Issue list - All issues in the epic

Unassigned Issues

Issues without an epic assignment appear in a separate “Unassigned Issues” section at the bottom of the Implementation.

Exporting to Platforms

Export selected issues to your project management platform:
1

Select Issues

Use the checkboxes to select the issues you want to export
2

Click Export

Click the Export button in the Implementation header
3

Select Platform

Choose your target platform:
  • GitHub Issues - Export as GitHub issues with milestone grouping
  • Linear - Export to Linear with project/cycle organization
  • Jira - Export to Jira with epic linking
  • Asana - Export to Asana with project/section organization
4

Configure Platform Options

Set platform-specific options based on your selection
5

Preview

Review the issues that will be created on the platform
6

Export

Click Export to create the issues on your platform
Export to Platform dialog showing platform selection

Platform-Specific Options

GitHub Issues

  • Repository - Target repository for issues
  • Labels - Apply labels to created issues
  • Create Milestones - Map epics to milestones

Linear

  • Team - Target Linear team
  • Project - Optional project assignment
  • Cycle - Optional cycle assignment

Jira

  • Project - Target Jira project
  • Issue Type - Default issue type for created issues
  • Create Epics - Map Implementation epics to Jira epics

Asana

  • Workspace - Target Asana workspace
  • Project - Target project
  • Create Sections - Map epics to project sections

Epic Mapping

Epics are mapped to platform concepts:
PlatformEpic Becomes
GitHubMilestone
LinearProject or Cycle
JiraEpic issue type
AsanaSection

After Export

Exported issues:
  • Are marked with an Exported badge in Kasava
  • Include links back to the platform issue
  • Show the platform issue identifier
  • Cannot be re-exported (prevents duplicates)

Code Context

When code context is enabled during generation, each issue includes AI-identified relevant information:

Affected Files

Files that will likely need modification:
src/api/auth.ts - Add authentication endpoint
src/middleware/session.ts - Update session handling
src/types/user.ts - Add new user fields
Existing patterns in your codebase to follow:
"React hook pattern" - see src/hooks/useAuth.ts
"API route handler" - see src/api/users.ts

Suggested Approach

AI-generated implementation hints based on your codebase patterns and conventions.

Implementation Status

Track progress through these statuses:
StatusDescriptionVisual
DraftNewly created, issues being reviewedFile icon with outline badge
In ProgressActively being worked onClock icon with primary badge
CompletedAll issues exported and resolvedCheckmark icon with success badge
ArchivedNo longer activeArchive icon with muted badge

Error Handling

If generation fails, you’ll see an error message with:
  • Error description - What went wrong
  • Retry option - For transient errors (network issues, AI service unavailable)
  • Suggestions - Tips to resolve the issue (e.g., try without code context)
Common errors include:
  • Network connectivity issues
  • AI service temporarily unavailable
  • Rate limiting (too many requests)
  • Repository access issues

Best Practices

Review before export - Always review AI-generated issues for accuracy and completeness before exporting to your project management platform.
Edit liberally - Adjust titles, descriptions, and priorities to match your team’s conventions and workflow.
Use epics strategically - Group work by component, skill required, or deployment phase for better organization.
Export incrementally - Export issues in batches to maintain focus and avoid overwhelming the backlog.
Select the right repository - Choose the primary repository to get the most relevant code context for your implementation.
Enable code context - Keep code context enabled to get AI-identified affected files and implementation hints.