kg CLI Usage Guide

Purpose: Comprehensive guide to all CLI commands, usage patterns, and command-line interface design.

Command Structure Overview

The kg CLI provides two complementary command styles (ADR-029):

Domain-Noun Commands (Primary): kg <noun> <verb> - e.g., kg job list, kg ontology delete "Name"
Unix-Verb Shortcuts (Secondary): kg <verb> <noun> - e.g., kg ls job, kg rm ontology "Name"

Both styles delegate to the same underlying commands and produce identical results.

Command Syntax

kg [global-options] <command> [subcommand] [options] [arguments]

Global Options

--api-url <url> - Override API base URL (default: http://localhost:8000)
--client-id <id> - Client ID for multi-tenancy (env: KG_CLIENT_ID)

Command Naming Convention

Commands use singular names with optional aliases for convenience:

kg job (alias: jobs) - Manage ingestion jobs
kg ontology (alias: onto) - Manage ontologies
kg database (alias: db) - Database operations
kg config (alias: cfg) - Configuration management

Examples:

kg job list          # Singular (primary)
kg jobs list         # Plural alias (backward compatible)
kg onto list         # Short alias

User-Configurable Aliases (ADR-029)

Users can define custom command aliases in their config:

kg config set aliases.cat '["bat"]'

This allows kg bat config to work alongside kg cat config for users with shell conflicts.

Unix Verb Router (ADR-029)

Unix-style shortcuts for common operations:

`kg ls <resource>`

List resources using Unix-familiar syntax.

Examples:

kg ls job              # → kg job list
kg ls ontology         # → kg ontology list
kg ls backup           # → kg admin list-backups

`kg cat <resource> [id]`

Display resource details (like Unix cat).

Examples:

kg cat config          # → kg config list (show all config)
kg cat config api_url  # → kg config get api_url
kg cat job             # → kg job list (show all jobs)
kg cat job job_xyz     # → kg job status job_xyz
kg cat concept abc-123 # → kg search details abc-123

`kg rm <resource> <id>`

Remove or delete resources.

Examples:

kg rm job job_abc          # → kg job cancel job_abc
kg rm ontology "My Docs"   # → kg ontology delete "My Docs"

`kg stat <resource> [id]`

Show status or statistics.

Examples:

kg stat database       # → kg database stats
kg stat job job_abc    # → kg job status job_abc

1. Health Command

Command: kg health

Purpose: Check API server health

States: - ✅ API healthy → display health info + API info - ❌ API unhealthy → error message, exit(1)

Flow:

kg health
  → GET /health
  → GET /info
  → Display results

2. Config Commands

Command: kg config <subcommand> (alias: cfg)

Unix Shortcuts:

kg cat config          # List all config (→ kg config list)
kg cat config api_url  # Show specific key (→ kg config get api_url)

2.1 `kg config get [key]`

Purpose: Get configuration value(s)

Options: - --json - Output as JSON

States: - No key → Show all config (formatted or JSON) - With key → Show specific key value - Key not found → Error, exit(1)

Flow:

kg config get
  → Load ~/.kg/config.json
  → Display all config

kg config get username
  → Load config
  → Get "username" value
  → Display

2.2 `kg config set <key> <value>`

Purpose: Set configuration value

Options: - --json - Force parse value as JSON - --string - Force treat value as string (no JSON parsing)

Auto-Detection: - Values starting with [ or { are automatically parsed as JSON - Values true/false are parsed as booleans - Numeric values are parsed as numbers - Other values are stored as strings

States: - Valid key/value → Set and confirm - Invalid JSON (with --json flag or auto-detected) → Error, exit(1)

Flow:

kg config set username alice
  → Load config
  → Auto-detect: string value
  → Set username = "alice"
  → Save config
  → Confirm

kg config set aliases.cat '["bat"]'
  → Auto-detect: JSON array (starts with [)
  → Parse as JSON
  → Set aliases.cat = ["bat"]
  → Save config

2.3 `kg config delete <key>`

Purpose: Delete configuration key

States: - Key exists → Delete and confirm - Key doesn't exist → Still succeeds (idempotent)

2.4 `kg config list`

Purpose: List all configuration

Options: - --json - Output as JSON

States: - Has config → Display formatted list - Empty config → Display empty structure

2.5 `kg config path`

Purpose: Show configuration file path

States: - Always succeeds → Display path

2.6 `kg config init`

Purpose: Initialize configuration file with defaults

Options: - -f, --force - Overwrite existing

States: - No config → Create with defaults - Config exists, no --force → Warning, exit(0) - Config exists, with --force → Overwrite with defaults

2.7 `kg config reset`

Purpose: Reset configuration to defaults

Options: - -y, --yes - Skip confirmation

States: - With --yes → Reset immediately - Without --yes → Prompt for confirmation - User confirms (y) → Reset - User cancels (n/other) → Exit(0)

2.8 `kg config auto-approve [value]`

Purpose: Enable/disable auto-approval of jobs (ADR-014)

States: - No value → Show current status - Value = true/on/yes/enable/enabled/1 → Enable auto-approve - Value = false/off/no/disable/disabled/0 → Disable auto-approve - Invalid value → Error, exit(1)

2.9 `kg config enable-mcp <tool>`

Purpose: Enable an MCP tool

2.10 `kg config disable-mcp <tool>`

Purpose: Disable an MCP tool

2.11 `kg config mcp [tool]`

Purpose: Show MCP tool configuration

Options: - --json - Output as JSON

2.12 `kg config update-secret`

Purpose: Authenticate and update API secret/key

Options: - -u, --username <username> - Username

States: - ⚠️ NOT IMPLEMENTED - Placeholder only

3. Ingest Commands

Command: kg ingest <subcommand>

3.1 `kg ingest file <path>`

Purpose: Ingest a document file

Options: - -o, --ontology <name> - REQUIRED - Ontology/collection name - -f, --force - Force re-ingestion even if duplicate (default: false) - -y, --yes - Auto-approve job, skip approval step (default: false) - --filename <name> - Override filename for tracking - --target-words <n> - Target words per chunk (default: 1000) - --overlap-words <n> - Overlap between chunks (default: 200) - --no-wait - Submit and exit (don't wait for completion)

States: 1. File validation - File not found → Error, exit(1) - File exists → Continue

Job submission
Duplicate detected → Show duplicate info, exit(0)
New job → Submit job
Wait behavior
With --wait (default) → Poll job with progress
With --no-wait → Return immediately with job ID

Flow:

kg ingest file doc.txt -o "My Ontology"
  → Validate file exists
  → POST /ingest (multipart/form-data)
  → Check for duplicate
    → If duplicate: Display info, exit
    → If new: Continue
  → If --wait (default):
      → Poll job with progress spinner
      → Display final result
  → If --no-wait:
      → Display job ID and exit immediately

3.2 `kg ingest text <text>`

Purpose: Ingest raw text

Options: - -o, --ontology <name> - REQUIRED - -f, --force - Force re-ingestion - -y, --yes - Auto-approve job - --filename <name> - Filename for tracking (default: "text_input") - --target-words <n> - Target words per chunk (default: 1000) - --no-wait - Submit and exit

States: - Same as kg ingest file above

4. Job Commands

Command: kg job <subcommand> (alias: jobs)

Unix Shortcuts:

kg cat job             # List all jobs (→ kg job list)
kg cat job <id>        # Show job status (→ kg job status <id>)
kg ls job              # List all jobs (→ kg job list)
kg stat job <id>       # Show job status (→ kg job status <id>)
kg rm job <id>         # Cancel job (→ kg job cancel <id>)

4.1 `kg job status <job-id>`

Purpose: Get job status

Options: - -w, --watch - Watch job until completion

States: - Job found, no --watch → Display status once - Job found, with --watch → Poll until completion - Job not found → Error, exit(1)

Flow:

kg job status job_123 --watch
  → GET /jobs/{job_id}
  → Poll every 2s until status ∈ {completed, failed, cancelled}
  → Display progress in real-time
  → Display final result

4.2 `kg job list`

Purpose: List recent jobs

Options: - -s, --status <status> - Filter by status - -c, --client <client-id> - Filter by client ID - -l, --limit <n> - Maximum jobs to return (default: 20)

States: - No jobs → "No jobs found" - Has jobs → Display table

Flow:

kg job list
  → GET /jobs?limit=20
  → Build table with columns:
      [Job ID, Client, Status, Ontology, Created, Progress]
  → Display table

4.3 `kg job list pending`

Purpose: List jobs awaiting approval

Options: - -c, --client <client-id> - -l, --limit <n> (default: 20)

4.4 `kg job list approved`

Purpose: List approved jobs (queued or processing)

4.5 `kg job list done`

Purpose: List completed jobs

4.6 `kg job list failed`

Purpose: List failed jobs

4.7 `kg job list cancelled`

Purpose: List cancelled jobs

4.8 `kg job approve <job-id-or-filter>`

Purpose: Approve a job or all jobs matching filter

Options: - -c, --client <client-id> - Filter by client ID (for batch operations)

States: - Starts with "job_" → Single job approval - Job found → Approve, display status - Job not found → Error, exit(1) - Filter keyword (pending, awaiting, approved, etc.) → Batch approval - Find jobs matching filter - Approve each job - Display summary (approved count, failed count)

4.9 `kg job cancel <job-id-or-filter>`

Purpose: Cancel a job or all jobs matching filter

Options: - -c, --client <client-id> - Filter by client ID

States: - Same pattern as approve above

5. Search Commands

Command: kg search <subcommand>

5.1 `kg search query <query>`

Purpose: Search for concepts using natural language

Arguments: - <query> - Search query text

Options: - -l, --limit <number> - Maximum results (default: 10) - --min-similarity <number> - Minimum similarity score 0.0-1.0 (default: 0.7)

States: - Results found → Display list with scores - No results → Display "Found 0 concepts"

Flow:

kg search query "recursive thinking"
  → POST /search/concepts
      { query: "recursive thinking", limit: 10, min_similarity: 0.7 }
  → Display results with:
      - Concept label
      - ID
      - Similarity score (colored)
      - Documents
      - Evidence count

5.2 `kg search details <concept-id>`

Purpose: Get detailed information about a concept

Arguments: - <concept-id> - Concept ID to retrieve

States: - Concept found → Display full details - Concept not found → Error, exit(1)

Flow:

kg search details concept_abc123
  → GET /concepts/{concept_id}
  → Display:
      - Label, ID, search terms
      - Evidence instances (quotes)
      - Relationships

5.3 `kg search related <concept-id>`

Purpose: Find concepts related through graph traversal

Arguments: - <concept-id> - Starting concept ID

Options: - -d, --depth <number> - Maximum traversal depth 1-5 (default: 2) - -t, --types <types...> - Filter by relationship types

States: - Related concepts found → Display grouped by distance - No related concepts → Display "Found 0 concepts"

Flow:

kg search related concept_abc123 --depth 3
  → POST /search/related
      { concept_id: "concept_abc123", max_depth: 3 }
  → Display results grouped by distance:
      Distance 1:
        • Concept A (path: IMPLIES)
      Distance 2:
        • Concept B (path: IMPLIES → SUPPORTS)

5.4 `kg search connect <from> <to>`

Purpose: Find shortest path between two concepts

Arguments: - <from> - Starting concept (ID or search phrase) - <to> - Target concept (ID or search phrase)

Options: - --max-hops <number> - Maximum path length (default: 5)

States: - Auto-detection: - Contains - or _ → Treat as concept ID - Otherwise → Treat as natural language query - Both IDs → Use ID-based search (POST /search/connect) - At least one query → Use search-based (POST /search/connect-by-search) - Path found → Display all paths with hops - No path found → "No connection found within N hops"

Flow:

kg search connect "linear thinking" "recursive depth"
  → Auto-detect: both are queries (no hyphens)
  → POST /search/connect-by-search
      { from_query: "linear thinking", to_query: "recursive depth", max_hops: 5 }
  → Match concepts
  → Find paths
  → Display paths with relationships

6. Database Commands

Command: kg database <subcommand> (alias: kg db)

6.1 `kg database stats`

Purpose: Show database statistics

States: - Connected → Display stats - Nodes (Concepts, Sources, Instances) - Relationships (Total, By Type) - Not connected → Error, exit(1)

Flow:

kg database stats
  → GET /database/stats
  → Display:
      Nodes:
        Concepts: 150
        Sources: 45
        Instances: 320
      Relationships:
        Total: 89
        By Type:
          IMPLIES: 30
          SUPPORTS: 25
          ...

6.2 `kg database info`

Purpose: Show database connection information

States: - Connected → Display connection details (URI, user, version, edition) - Not connected → Display error details

6.3 `kg database health`

Purpose: Check database health and connectivity

States: - Healthy → Display "✓ HEALTHY" - Degraded → Display "⚠ DEGRADED" with warnings - Unhealthy → Display "✗ UNHEALTHY" with errors

7. Ontology Commands

Command: kg ontology <subcommand> (alias: onto)

7.1 `kg ontology list`

Purpose: List all ontologies

States: - No ontologies → "⚠ No ontologies found" - Has ontologies → Display list with stats - Files, Chunks, Concepts per ontology

7.2 `kg ontology info <name>`

Purpose: Get detailed information about an ontology

Arguments: - <name> - Ontology name

States: - Ontology found → Display full info - Statistics (files, chunks, concepts, evidence, relationships) - File list - Ontology not found → Error, exit(1)

7.3 `kg ontology files <name>`

Purpose: List files in an ontology

Arguments: - <name> - Ontology name

States: - Files found → Display list with counts (chunks, concepts per file) - No files → "No files found"

7.4 `kg ontology delete <name>`

Purpose: Delete an ontology and all its data

Arguments: - <name> - Ontology name

Options: - -f, --force - Skip confirmation and force deletion

States: - No --force → Display warning and require --force flag, exit(0) - With --force → Delete and display results - Sources deleted count - Orphaned concepts cleaned count

Flow:

kg ontology delete "Test Ontology"
  → Display warning
  → Exit (requires --force)

kg ontology delete "Test Ontology" --force
  → DELETE /ontology/{name}?force=true
  → Display deletion results

8. Admin Commands

Command: kg admin <subcommand>

8.1 `kg admin status`

Purpose: Show system status (Docker, database, environment)

States: - All components healthy → Display all green ✓ - Some components unhealthy → Display mixed status

Flow:

kg admin status
  → GET /admin/system-status
  → Display:
      Docker:
        ✓ PostgreSQL container running
      Database Connection:
        ✓ Connected to PostgreSQL + AGE
      Database Statistics:
        Concepts: 150, Sources: 45, ...
      Python Environment:
        ✓ Virtual environment exists
      Configuration:
        ✓ .env file exists
        ✓ ANTHROPIC_API_KEY: configured

8.2 `kg admin backup`

Purpose: Create a database backup

Options: - --type <type> - Backup type: "full" or "ontology" - --ontology <name> - Ontology name (required if type is ontology) - --output <filename> - Custom output filename

States: - Interactive mode (no options): - Prompt: "1) Full database backup" or "2) Specific ontology backup" - If ontology: Prompt for ontology name - Non-interactive mode: - Validate options - Download backup

Flow:

kg admin backup --type full
  → POST /admin/backup
      { backup_type: "full" }
  → Download with progress bar
  → Save to configured backup directory
  → Display results (filename, path, size)

8.3 `kg admin list-backups`

Purpose: List available backup files from configured directory

States: - Backup dir doesn't exist → Display message - No backups → Display "No backups found" - Has backups → Display list (newest first) - Filename, size, created date

8.4 `kg admin restore`

Purpose: Restore a database backup (requires authentication)

Options: - --file <name> - Backup filename (from configured directory) - --path <path> - Custom backup file path - --overwrite - Overwrite existing data (default: false) - --deps <action> - Handle external dependencies: prune, stitch, defer (default: prune)

States: 1. File selection: - --path → Use custom path - --file → Use configured directory + filename - Neither → Interactive selection from configured directory

File validation:
File not found → Error, exit(1)
File exists → Continue
Authentication:
Get username from config or prompt
Prompt for password (hidden input)
Username not configured → Error, exit(1)
No password → Error, exit(1)
Upload backup:
Upload with progress bar
Display backup stats if available
Display integrity warnings if any
Restore job tracking (ADR-018):
Try SSE streaming for real-time progress
Fall back to polling if SSE fails
Display multi-line progress bars for stages:
- Creating checkpoint backup
- Loading backup file
- Restoring concepts
- Restoring sources
- Restoring instances
- Restoring relationships
Final status:
Completed → Display restore statistics
Failed → Display error, check rollback

Flow:

kg admin restore --file backup_2024-10-09.json
  → Validate file exists
  → Prompt for authentication
  → POST /admin/restore (multipart/form-data)
      + username, password, overwrite, deps
  → Upload with progress
  → Get job_id
  → Track job with SSE (or polling fallback)
  → Display multi-line progress
  → Display final results

8.5 `kg admin reset`

Purpose: Reset database - DESTRUCTIVE (requires authentication)

Options: - --no-logs - Do not clear log files - --no-checkpoints - Do not clear checkpoint files

States: 1. Confirmation: - Prompt user to type "yes" to confirm - User types "yes" → Continue - User types anything else → Exit(0)

Authentication:
Get username from config
Prompt for password
Validate inputs
Reset operation:
Displays schema validation results
Displays warnings if any

Flow:

kg admin reset
  → Display warnings
  → Prompt: Type "yes" to confirm
  → POST /admin/reset
      { username, password, confirm: true, clear_logs, clear_checkpoints }
  → Display schema validation
  → Display warnings

8.6 `kg admin scheduler status`

Purpose: Show job scheduler status and configuration

States: - Scheduler running → Display "✓ Running" - Scheduler not running → Display "✗ Not running"

Flow:

kg admin scheduler status
  → GET /admin/scheduler/status
  → Display:
      Scheduler: ✓ Running
      Configuration:
        Cleanup Interval: 3600s (1.0h)
        Approval Timeout: 24h
        ...
      Job Statistics:
        awaiting_approval: 2
        completed: 50
        ...

8.7 `kg admin scheduler cleanup`

Purpose: Manually trigger scheduler cleanup

Flow:

kg admin scheduler cleanup
  → POST /admin/scheduler/cleanup
  → Display cleanup results

9. Authentication Commands

Note: Detailed authentication documentation is in 01-AUTHENTICATION.md (ADR-054: OAuth 2.0)

The Knowledge Graph System uses OAuth 2.0 client credentials for authentication. All authenticated commands automatically handle token refresh.

9.1 `kg login`

Command: kg login [options]

Purpose: Authenticate and create personal OAuth client credentials

Options: - -u, --username <username> - Username (will prompt if not provided)

Flow:

kg login
  → Prompt for username (if not provided or saved)
  → Prompt for password (hidden input)
  → POST /auth/oauth/clients/personal
  → Save OAuth client credentials to ~/.kg/config.json
  → Display success message with client info

What it does: - Authenticates with username/password - Creates a long-lived OAuth client (client_id + client_secret) - Stores OAuth credentials locally (NOT the password) - Future API requests use OAuth client credentials grant

Example:

kg login

# Output:
# Username: admin
# Password: ********
#
# ✓ Creating personal OAuth client credentials...
# ✓ Login successful
#
# Logged in as: admin (role: admin)
# OAuth Client: kg-cli-admin-20251102
# Scopes: read:*, write:*

9.2 `kg logout`

Command: kg logout [options]

Purpose: Revoke OAuth client and clear credentials

Options: - --forget - Also forget saved username

Flow:

kg logout
  → Get OAuth client credentials from config
  → DELETE /auth/oauth/clients/personal/{client_id}
  → Remove OAuth credentials from config
  → Display success message

What it does: - Revokes the OAuth client at the server (invalidates credentials) - Clears local OAuth credentials from config - Optionally clears saved username (with --forget)

Example:

kg logout

# Output:
# ✓ OAuth client revoked
# ✓ Logged out successfully

9.3 `kg oauth` - OAuth Client Management

Command: kg oauth <subcommand>

Purpose: Manage personal OAuth clients (list, create for MCP, revoke)

9.3.1 `kg oauth clients` (alias: `kg oauth list`)

Purpose: List all personal OAuth clients

Flow:

kg oauth clients
  → GET /auth/oauth/clients/personal
  → Display table of OAuth clients

Example:

kg oauth clients

# Output:
# Personal OAuth Clients
#
# Client ID                 Name                 Scopes          Created              Status
# ────────────────────────────────────────────────────────────────────────────────────────────
# kg-cli-admin-20251102     kg CLI (admin)       read:*, write:* 2 hours ago          ✓ Active
# kg-mcp-server-admin       kg MCP Server (a...  read:*, write:* 1 day ago            ✓ Active

9.3.2 `kg oauth create-mcp`

Purpose: Create OAuth client for MCP server and display ready-to-paste config

Options: - --name <name> - Custom client name (default: "kg MCP Server (username)")

Flow:

kg oauth create-mcp
  → POST /auth/oauth/clients/personal/new
  → Display OAuth credentials
  → Display Claude Desktop config (ready to paste)
  → Display claude CLI command

Example:

kg oauth create-mcp

# Output:
# 🔐 Creating OAuth client for MCP server...
#
# ✅ OAuth client created successfully!
#
# ════════════════════════════════════════════════════════════════════════════════
# CLAUDE DESKTOP CONFIG
# ════════════════════════════════════════════════════════════════════════════════
#
# Add this to your Claude Desktop config:
#
#   "knowledge-graph": {
#     "command": "kg-mcp-server",
#     "env": {
#       "KG_OAUTH_CLIENT_ID": "kg-mcp-server-admin-20251102",
#       "KG_OAUTH_CLIENT_SECRET": "oauth_secret_abc123...",
#       "KG_API_URL": "http://localhost:8000"
#     }
#   }
#
# ⚠️  IMPORTANT:
#   • Keep these credentials secure!
#   • Client secret is shown only once
#   • To revoke: kg oauth revoke kg-mcp-server-admin-20251102

9.3.3 `kg oauth revoke <client-id>`

Purpose: Revoke an OAuth client

Options: - --force - Force revocation even if it's your current CLI client

Flow:

kg oauth revoke <client-id>
  → Check if revoking current CLI client
  → DELETE /auth/oauth/clients/personal/{client_id}
  → Display success message
  → If current CLI client: clear config and logout

Example:

# Revoke MCP server client
kg oauth revoke kg-mcp-server-admin-20251102

# Output:
# 🗑️  Revoking OAuth client kg-mcp-server-admin-20251102...
#
# ✅ OAuth client revoked successfully!

# Try to revoke current CLI client (protected)
kg oauth revoke kg-cli-admin-20251102

# Output:
# ⚠️  Warning: This is your current CLI OAuth client
#    Client ID: kg-cli-admin-20251102
#    Revoking this will log you out.
#
#    To proceed, use: kg oauth revoke kg-cli-admin-20251102 --force
#    Or use: kg logout

kg CLI Usage Guide

Command Structure Overview

Command Syntax

Global Options

Command Naming Convention

User-Configurable Aliases (ADR-029)

Unix Verb Router (ADR-029)

kg ls <resource>

kg cat <resource> [id]

kg rm <resource> <id>

kg stat <resource> [id]

1. Health Command

2. Config Commands

2.1 kg config get [key]

2.2 kg config set <key> <value>

2.3 kg config delete <key>

2.4 kg config list

2.5 kg config path

2.6 kg config init

2.7 kg config reset

2.8 kg config auto-approve [value]

2.9 kg config enable-mcp <tool>

2.10 kg config disable-mcp <tool>

2.11 kg config mcp [tool]

2.12 kg config update-secret

3. Ingest Commands

3.1 kg ingest file <path>

3.2 kg ingest text <text>

4. Job Commands

4.1 kg job status <job-id>

4.2 kg job list

4.3 kg job list pending

4.4 kg job list approved

4.5 kg job list done

4.6 kg job list failed

4.7 kg job list cancelled

4.8 kg job approve <job-id-or-filter>

4.9 kg job cancel <job-id-or-filter>

5. Search Commands

5.1 kg search query <query>

5.2 kg search details <concept-id>

5.3 kg search related <concept-id>

5.4 kg search connect <from> <to>

6. Database Commands

6.1 kg database stats

6.2 kg database info

6.3 kg database health

7. Ontology Commands

7.1 kg ontology list

7.2 kg ontology info <name>

7.3 kg ontology files <name>

7.4 kg ontology delete <name>

8. Admin Commands

8.1 kg admin status

8.2 kg admin backup

8.3 kg admin list-backups

8.4 kg admin restore

8.5 kg admin reset

8.6 kg admin scheduler status

8.7 kg admin scheduler cleanup

9. Authentication Commands

9.1 kg login

9.2 kg logout

9.3 kg oauth - OAuth Client Management

9.3.1 kg oauth clients (alias: kg oauth list)

9.3.2 kg oauth create-mcp

9.3.3 kg oauth revoke <client-id>

`kg ls <resource>`

`kg cat <resource> [id]`

`kg rm <resource> <id>`

`kg stat <resource> [id]`

2.1 `kg config get [key]`

2.2 `kg config set <key> <value>`

2.3 `kg config delete <key>`

2.4 `kg config list`

2.5 `kg config path`

2.6 `kg config init`

2.7 `kg config reset`

2.8 `kg config auto-approve [value]`

2.9 `kg config enable-mcp <tool>`

2.10 `kg config disable-mcp <tool>`

2.11 `kg config mcp [tool]`

2.12 `kg config update-secret`

3.1 `kg ingest file <path>`

3.2 `kg ingest text <text>`

4.1 `kg job status <job-id>`

4.2 `kg job list`

4.3 `kg job list pending`

4.4 `kg job list approved`

4.5 `kg job list done`

4.6 `kg job list failed`

4.7 `kg job list cancelled`

4.8 `kg job approve <job-id-or-filter>`

4.9 `kg job cancel <job-id-or-filter>`

5.1 `kg search query <query>`

5.2 `kg search details <concept-id>`

5.3 `kg search related <concept-id>`

5.4 `kg search connect <from> <to>`

6.1 `kg database stats`

6.2 `kg database info`

6.3 `kg database health`

7.1 `kg ontology list`

7.2 `kg ontology info <name>`

7.3 `kg ontology files <name>`

7.4 `kg ontology delete <name>`

8.1 `kg admin status`

8.2 `kg admin backup`

8.3 `kg admin list-backups`

8.4 `kg admin restore`

8.5 `kg admin reset`

8.6 `kg admin scheduler status`

8.7 `kg admin scheduler cleanup`

9.1 `kg login`

9.2 `kg logout`

9.3 `kg oauth` - OAuth Client Management

9.3.1 `kg oauth clients` (alias: `kg oauth list`)

9.3.2 `kg oauth create-mcp`

9.3.3 `kg oauth revoke <client-id>`