### Initialize Kelar Framework

Source: https://context7.com/zeative/kelar/llms.txt

Installs and configures the Kelar CLI tool within a project. The process includes selecting an AI agent, defining the installation scope, and configuring behavioral preferences.

```bash
# Install KELAR with interactive setup
npx kelar-cli@latest init

# After installation, map your codebase
/kelar:map
```

--------------------------------

### KELAR Specialized Agents Initialization (Bash)

Source: https://context7.com/zeative/kelar/llms.txt

Demonstrates how KELAR spawns specialized sub-agents using the Task() function in workflows. Each agent has a specific role, such as planning, execution, research, debugging, verification, and repair. The prompts define the agent's persona and instructions, including files to read and context.

```bash
# Agents are spawned via Task() in workflows:

# kelar-planner - Creates XML task plans
Task(
  subagent_type="general-purpose",
  prompt="""You are kelar-planner. Create XML plan for [feature].
  <files_to_read>
  AGENTS.md
  .kelar/state/STATE.md
  .kelar/state/PATTERNS.md
  </files_to_read>"""
)

# kelar-executor - Implements single tasks from plans
Task(
  subagent_type="general-purpose",
  prompt="""You are kelar-executor. Implement this task exactly as specified.
  <task>[XML task from plan]</task>
  <files_to_read>
  AGENTS.md
  [target file]
  </files_to_read>"""
)

# kelar-researcher - Investigates libraries and codebase patterns
Task(
  subagent_type="general-purpose",
  prompt="""You are kelar-researcher. Research [topic].
  Check memory: $(node .kelar/kelar-tools.cjs memory search "[topic]")
  Write findings to .kelar/research/[slug]-research.md"""
)

# kelar-debugger - Traces bugs 3+ levels deep to root cause
Task(
  subagent_type="general-purpose",
  prompt="""You are kelar-debugger. Analyze this error.
  [error details]
  Trace 3 levels deep. Produce root cause and 3 fix options."""
)

# kelar-verifier - Confirms goals achieved, not just code compiled
Task(
  subagent_type="general-purpose",
  prompt="""You are kelar-verifier. Verify [feature] implementation.
  Goal: [expected outcome]
  Check: existence, compilation, tests, goal achievement, scope."""
)

# kelar-repair - Autonomous recovery with retry, decompose, prune
Task(
  subagent_type="general-purpose",
  prompt="""You are kelar-repair. Recover from failed task.
  Failed task: [task XML]
  Repair budget: 2 attempts"""
)

# kelar-plan-checker - Validates plans before execution
# kelar-ui-designer - Creates design contracts with all 8 component states
# kelar-codebase-mapper - Full architecture analysis

```

--------------------------------

### Display Project Status Dashboard

Source: https://context7.com/zeative/kelar/llms.txt

Provides a high-level overview of the project's current state, including active tasks, progress, technical debt, and overall system health.

```bash
# Get current project status
/kelar:status
```

--------------------------------

### Map Codebase Architecture

Source: https://context7.com/zeative/kelar/llms.txt

Scans the project to document architecture, naming conventions, and technical debt. It generates state files in the .kelar directory to inform future AI operations.

```bash
# Execute the map workflow
/kelar:map
```

--------------------------------

### XML Plan Structure for KELAR Tasks

Source: https://context7.com/zeative/kelar/llms.txt

Defines the structure of XML plans used by KELAR to organize atomic, verifiable tasks into parallel and sequential waves. Includes metadata, wave definitions, task details (ID, title, file, action, verification, completion status, dependencies), out-of-scope items, and risks.

```xml
<!-- Example plan structure: .kelar/plans/feature-plan.xml -->
<kelar_plan>
  <meta>
    <feature>User Authentication</feature>
    <goal>Users can sign in with email/password and receive JWT token</goal>
    <wave_count>3</wave_count>
    <created>2024-01-15T10:00:00Z</created>
  </meta>

  <wave number="1" title="Foundation" parallel="true">
    <task id="1.1">
      <title>Define authentication types</title>
      <file>src/types/auth.ts</file>
      <action>
        Create User interface with id, email, passwordHash fields.
        Create AuthResult type with success: boolean, token?: string, error?: string.
        Export both types.
      </action>
      <verify>tsc --noEmit passes with no errors</verify>
      <done>Types are importable from src/types/auth.ts</done>
      <depends_on></depends_on>
    </task>

    <task id="1.2">
      <title>Add JWT utilities</title>
      <file>src/utils/jwt.ts</file>
      <action>
        Add generateToken(userId: string): string using jsonwebtoken.
        Add verifyToken(token: string): {userId: string} | null.
        Use JWT_SECRET from env.
      </action>
      <verify>Functions exist and compile</verify>
      <done>JWT utilities generate and verify tokens correctly</done>
      <depends_on></depends_on>
    </task>
  </wave>

  <wave number="2" title="Core Logic" parallel="false">
    <task id="2.1">
      <title>Create AuthService</title>
      <file>src/services/auth.ts</file>
      <action>
        Add login(email, password): Promise<AuthResult> method.
        Use bcrypt.compare for password verification.
        Return token on success, throw AuthError on failure.
        Follow existing service patterns in src/services/user.ts.
      </action>
      <verify>AuthService.login compiles and handles both success/failure</verify>
      <done>AuthService returns JWT on valid credentials, throws on invalid</done>
      <depends_on>1.1, 1.2</depends_on>
    </task>
  </wave>

  <out_of_scope>
    Password reset flow
    OAuth providers
    Rate limiting
  </out_of_scope>

  <risks>
    <risk level="LOW">JWT secret must be set in environment</risk>
  </risks>
</kelar_plan>
```

--------------------------------

### Implement Features with Multi-Agent Pipeline

Source: https://context7.com/zeative/kelar/llms.txt

Orchestrates a multi-step feature implementation process including research, planning, human validation, and parallel task execution.

```bash
# Start a feature implementation
/kelar:feature "Add user authentication with JWT"
```

--------------------------------

### KELAR CLI Tool Operations

Source: https://github.com/zeative/kelar/blob/main/templates/AGENTS.md

A collection of commands for the KELAR CLI utility used to manage session state, task logging, memory, version control, and technical debt. These commands should be used in place of raw shell commands to ensure consistency across the framework.

```bash
# Session start
node .kelar/kelar-tools.cjs health
node .kelar/kelar-tools.cjs state snapshot
node .kelar/kelar-tools.cjs handoff read
node .kelar/kelar-tools.cjs memory search "[current topic]"

# Task logging
node .kelar/kelar-tools.cjs tasks log start "Task: [name] | Files: [list]"
node .kelar/kelar-tools.cjs tasks log done "Task [id]: [what was done]"
node .kelar/kelar-tools.cjs tasks log pause "[task id]" "[exact next step]"

# Memory operations
node .kelar/kelar-tools.cjs memory save technical "[title]" "[content]"
node .kelar/kelar-tools.cjs memory search "[query]"

# Git operations
node .kelar/kelar-tools.cjs git status
node .kelar/kelar-tools.cjs git commit "feat(kelar): [message]"
node .kelar/kelar-tools.cjs git checkpoint

# Debt management
node .kelar/kelar-tools.cjs debt add "[file]" "[issue]" "MEDIUM"

# Pattern management
node .kelar/kelar-tools.cjs patterns get "[category]"
node .kelar/kelar-tools.cjs patterns set "[category]" "[pattern]"

# Plan operations
node .kelar/kelar-tools.cjs plan validate .kelar/plans/[name]-plan.xml
node .kelar/kelar-tools.cjs plan wave .kelar/plans/[name]-plan.xml 1
```

--------------------------------

### KELAR Task Completion Template

Source: https://github.com/zeative/kelar/blob/main/templates/AGENTS.md

The standardized format required for reporting the completion of any task within the KELAR framework. This ensures traceability and quality assurance for all agent actions.

```text
KELAR TASK COMPLETE
────────────────────
Done     : [task]
Files    : [list with summaries]
Result   : [what user can now do]
Agents   : [which agents ran]
Quality  : [gates passed / issues]
Next     : [suggestion]
```

--------------------------------

### Perform Root Cause Bug Fixing

Source: https://context7.com/zeative/kelar/llms.txt

Uses specialized debugging agents to trace errors to their source rather than applying superficial patches. It presents multiple fix strategies for user selection before execution.

```bash
# Start a bug fix with error details
/kelar:fix "TypeError: Cannot read property 'id' of undefined at UserService.ts:45"
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.