### Run Dorothy in Web Browser (Development)

Source: https://github.com/charlie85270/dorothy/blob/main/README.md

Installs dependencies and starts the Dorothy web development server. Access the application at http://localhost:3000. Note that agent management and terminal features require the Electron app.

```bash
npm install
npm run dev
```

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

### Build Dorothy from Source (Development)

Source: https://github.com/charlie85270/dorothy/blob/main/README.md

Installs dependencies and starts the Dorothy application in development mode. Requires Node.js and npm.

```bash
git clone https://github.com/Charlie85270/Dorothy.git
cd Dorothy/app/dorothy
npm install
npx @electron/rebuild        # Rebuild native modules for Electron
npm run electron:dev          # Development mode
```

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

### Example: Store Learning about Codebase Pattern

Source: https://github.com/charlie85270/dorothy/blob/main/skills/remember.md

This example shows how to store learned information about a codebase pattern. Replace {{AGENT_ID}} with your agent's ID.

```bash
curl -s -X POST http://127.0.0.1:31415/api/memory/remember \
  -H "Content-Type: application/json" \
  -d '{"agent_id": "{{AGENT_ID}}", "content": "Project uses repository pattern for data access", "type": "learning"}'
```

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

### Example: Multi-Agent Task - Narrate Parallel Start

Source: https://github.com/charlie85270/dorothy/blob/main/electron/resources/telegram-instructions.md

Inform the user when multiple agents are being started simultaneously for different parts of the task.

```python
send_telegram("Starting deploy-agent on deployment and docs-agent on documentation...")
```

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

### Install and Build MCP Orchestrator

Source: https://github.com/charlie85270/dorothy/blob/main/mcp-orchestrator/README.md

Steps to install dependencies and build the MCP orchestrator project.

```bash
cd mcp-orchestrator
npm install
npm run build
```

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

### Example: Store User Name Preference

Source: https://github.com/charlie85270/dorothy/blob/main/skills/remember.md

This example demonstrates how to store a user's name as a preference. Replace {{AGENT_ID}} with your agent's ID.

```bash
curl -s -X POST http://127.0.0.1:31415/api/memory/remember \
  -H "Content-Type: application/json" \
  -d '{"agent_id": "{{AGENT_ID}}", "content": "User name is Charlie", "type": "preference"}'
```

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

### Narrate Starting Agent

Source: https://github.com/charlie85270/dorothy/blob/main/electron/resources/telegram-instructions.md

Before initiating an agent, inform the user about the agent being started and the task it will perform.

```python
send_telegram("Starting [agent name] on [task]...")
```

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

### Run Next.js Development Server

Source: https://github.com/charlie85270/dorothy/blob/main/landing/README.md

Commands to start the Next.js development server using different package managers.

```bash
npm run dev
# or
yarn dev
# or
pnpm dev
# or
bun dev
```

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

### Cron Schedule Examples

Source: https://github.com/charlie85270/dorothy/blob/main/README.md

Provides common examples of cron expressions for scheduling tasks. These illustrate various frequencies and times.

```text
Expression | Schedule |
-----------|----------|
`0 9 * * *` | Daily at 9:00 AM |
`0 9 * * 1-5` | Weekdays at 9:00 AM |
`*/15 * * * *` | Every 15 minutes |
`0 */2 * * *` | Every 2 hours |
`30 14 * * 1` | Mondays at 2:30 PM |
```

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

### start_agent

Source: https://github.com/charlie85270/dorothy/blob/main/mcp-orchestrator/README.md

Starts a previously created agent with a given task or prompt.

```APIDOC
## start_agent

### Description
Start an agent with a specific task/prompt.

### Method
POST

### Endpoint
/agents/{id}/start

### Parameters
#### Path Parameters
- **id** (string) - Required - The unique identifier of the agent to start.
#### Request Body
- **prompt** (string) - Required - The task or instruction for the agent.
- **model** (string) - Optional - The specific model to use for the agent's task.

### Response
#### Success Response (200)
- **status** (string) - Indicates the agent has started successfully.

#### Response Example
{
  "status": "started"
}
```

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

### Example: Simple Task - Narrate Action

Source: https://github.com/charlie85270/dorothy/blob/main/electron/resources/telegram-instructions.md

After identifying the correct agent, inform the user that the task is being initiated and may take some time.

```python
send_telegram("Found the test agent. Running auth module tests now... This may take a minute.")
```

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

### Twitter/X World Generation Example

Source: https://github.com/charlie85270/dorothy/blob/main/skills/world-builder/SKILL.md

An example workflow for generating a world based on a Twitter/X user. It outlines the sequence of tool calls for data gathering and zone creation.

```text
Input: "Build a world about @marcZeller"

Step 1: list_sprites → pick sprites matching the theme (1 call)
Step 2: socialdata → get_profile("marcZeller") + search_tweets("from:marcZeller") (2 calls)
Step 3: web search → "Marc Zeller Aave governance" + "Marc Zeller DeFi" (2 calls)
Step 4: create_zone with NPCs quoting real tweets, buildings as projects, interiors with lore (1 call)
DONE. Total: 6 tool calls.
```

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

### Example: Simple Task - Initial Acknowledgment

Source: https://github.com/charlie85270/dorothy/blob/main/electron/resources/telegram-instructions.md

The first step in handling a user's request is to acknowledge receipt and state the intention to find the appropriate agent.

```python
send_telegram("On it! Let me find the right agent for this.")
```

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

### Install Claude Code CLI

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/source-content.md

Install the Claude Code command-line interface globally using npm. This is the first step to setting up your Claude Code environment.

```bash
npm install -g @anthropic-ai/claude-code
```

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

### SessionStart Hook Script Example

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/source-content.md

A bash script for the 'SessionStart' hook that loads project context by displaying recent git commits, the content of a '.claude-context.md' file if it exists, the current git branch, and the working directory.

```bash
#!/bin/bash
# ~/.claude/hooks/session-start.sh

# Load recent git activity
echo "Recent commits:"
git log --oneline -5

# Load project context
if [ -f ".claude-context.md" ]; then
  cat .claude-context.md
fi

# Show current branch and status
echo "Current branch: $(git branch --show-current)"
echo "Working directory: $(pwd)"
```

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

### Example: Multi-Agent Task - Initial Acknowledgment

Source: https://github.com/charlie85270/dorothy/blob/main/electron/resources/telegram-instructions.md

For complex tasks involving multiple agents, provide an initial acknowledgment and inform the user about the parallel coordination.

```python
send_telegram("Got it — I'll coordinate the deploy and docs update in parallel.")
```

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

### Example: Simple Task - Final Confirmation

Source: https://github.com/charlie85270/dorothy/blob/main/electron/resources/telegram-instructions.md

Upon successful completion of a task, provide a detailed summary of the results to the user.

```python
send_telegram("Tests completed! Results: 15 passed, 0 failed. All auth tests passing. ✅")
```

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

### Interior Structure Example

Source: https://github.com/charlie85270/dorothy/blob/main/skills/world-builder/SKILL.md

Defines the structure for an enterable building interior, including background image and NPC placement.

```json
{
  "buildingId": "shop-1",
  "backgroundImage": "/pokemon/interior/sprite_3.png",
  "npcs": [
    {
      "id": "shopkeeper",
      "name": "Shopkeeper",
      "x": 5, "y": 2,
      "direction": "down",
      "spritePath": "/pokemon/pnj/NPC_Shopkeeper.png",
      "dialogue": ["Welcome to my shop!", "Everything is overpriced, just like real life."]
    }
  ]
}
```

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

### Example: Multi-Agent Task - Waiting for Second Agent

Source: https://github.com/charlie85270/dorothy/blob/main/electron/resources/telegram-instructions.md

After one agent completes, inform the user and indicate that you are now waiting for the next agent.

```python
send_telegram("Deploy finished successfully! Now waiting on docs...")
```

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

### Development Scripts

Source: https://github.com/charlie85270/dorothy/blob/main/README.md

Run these npm scripts for local development and building the project. Includes commands for starting development servers, building for production, and packaging the Electron app.

```bash
npm run dev              # Next.js dev server
npm run electron:dev     # Electron + Next.js concurrent dev mode
npm run build            # Next.js production build
npm run electron:build   # Distributable Electron app (DMG)
npm run electron:pack    # Electron directory package
npm run lint             # ESLint
```

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

### Example: Multi-Agent Task - Waiting for First Agent

Source: https://github.com/charlie85270/dorothy/blob/main/electron/resources/telegram-instructions.md

When coordinating multiple agents, update the user on the progress, specifically mentioning waiting for the first agent to complete.

```python
send_telegram("Both agents are working. Waiting for deploy to finish first...")
```

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

### Example: Multi-Agent Task - Final Confirmation

Source: https://github.com/charlie85270/dorothy/blob/main/electron/resources/telegram-instructions.md

Conclude multi-agent tasks by sending a comprehensive final confirmation that includes a summary of all completed actions.

```python
send_telegram("All done! Deploy completed and docs are updated. Here's the summary: [details]")
```

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

### Example Claude Code CLI Command

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/ai-dev-practice/prompts/10-slide-claude-code-intro.md

Use this command to integrate Claude into your development workflow for tasks like adding authentication to a user service.

```bash
claude "Add authentication to user service"
```

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

### User Prompt Submit Hook Example

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/source-content.md

This bash script runs before each user prompt. It injects a timestamp, displays the latest test run summary if available, and searches for relevant memories using 'claude-mem'.

```bash
#!/bin/bash
# ~/.claude/hooks/user-prompt-submit.sh

# Add timestamp
echo "Request time: $(date '+%Y-%m-%d %H:%M:%S')"

# Check if tests passing
if [ -f "test-results.json" ]; then
  echo "Latest test run: $(jq -r '.summary' test-results.json)"
fi

# Load relevant memories
if command -v claude-mem &> /dev/null; then
  claude-mem search "$USER_PROMPT" --limit 3
fi
```

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

### Configure Claude Code Hooks

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/source-content.md

Defines shell scripts to run at specific Claude workflow events. This example configures the 'SessionStart' hook to execute a custom script.

```json
{
  "hooks": {
    "SessionStart": {
      "command": "~/.claude/hooks/session-start.sh"
    }
  }
}
```

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

### Tool Call After Hook Example

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/source-content.md

This bash script runs after a tool call. It logs whether the tool succeeded or failed to a metrics file, helping to track AI performance.

```bash
#!/bin/bash
# ~/.claude/hooks/tool-call-after.sh

TOOL_NAME="$1"
SUCCESS="$2"  # true/false

# Track metrics
if [ "$SUCCESS" == "true" ]; then
  echo "✓ $TOOL_NAME succeeded" >> ~/.claude/metrics.txt
else
  echo "✗ $TOOL_NAME failed" >> ~/.claude/metrics.txt
fi
```

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

### Shell Script for SessionStart Hook

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/ai-dev-practice/prompts/12-slide-hook-sessionstart.md

This bash script is executed when a Claude session starts. It displays recent git commits, the content of a .claude-context.md file, and the current git branch, injecting this information as context for the AI.

```bash
#!/bin/bash
echo "Recent commits:"
git log --oneline -5
cat .claude-context.md
echo "Branch: $(git branch --show-current)"
```

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

### Create JIRA Issue Processor Automation

Source: https://github.com/charlie85270/dorothy/blob/main/README.md

Set up an automation to poll JIRA for specific issues based on JQL and trigger an agent to process them. This example configures polling, agent prompt with JIRA variables, and directs output to JIRA comments, transitions, and Telegram.

```javascript
create_automation({
  name: "JIRA Task Agent",
  sourceType: "jira",
  sourceConfig: '{"projectKeys": ["PROJ"], "jql": "status = Open"}',
  scheduleMinutes: 5,
  agentEnabled: true,
  agentPrompt: "Work on JIRA issue {{key}}: {{summary}}. Description: {{body}}. Priority: {{priority}}.",
  agentProjectPath: "/path/to/project",
  outputJiraComment: true,
  outputJiraTransition: true,
  outputTelegram: true
})
```

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

### Tool Call Before Hook Example

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/source-content.md

This bash script executes before a tool call. It logs the tool name and parameters to a file and includes a validation check to warn about potentially destructive commands like 'rm -rf'.

```bash
#!/bin/bash
# ~/.claude/hooks/tool-call-before.sh

TOOL_NAME="$1"
TOOL_PARAMS="$2"

# Log the action
echo "[$(date)] Tool: $TOOL_NAME" >> ~/.claude/tool-log.txt

# Validate dangerous operations
if [[ "$TOOL_NAME" == "Bash" ]] && [[ "$TOOL_PARAMS" =~ "rm -rf" ]]; then
  echo "⚠️  WARNING: Destructive command detected"
  echo "Command: $TOOL_PARAMS"
  # Could exit 1 to block execution
fi
```

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

### Get Memory Statistics using cURL

Source: https://github.com/charlie85270/dorothy/blob/main/skills/remember.md

Use this command to retrieve statistics about the memory system.

```bash
curl -s http://127.0.0.1:31415/api/memory/stats
```

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

### Build Dorothy from Source (Production)

Source: https://github.com/charlie85270/dorothy/blob/main/README.md

Builds a production-ready version of the Dorothy application. This command generates distributable files in the 'release/' directory.

```bash
git clone https://github.com/Charlie85270/Dorothy.git
cd Dorothy/app/dorothy
npm install
npx @electron/rebuild        # Rebuild native modules for Electron
npm run electron:build        # Production build (DMG)
```

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

### Basic Telegram Update

Source: https://github.com/charlie85270/dorothy/blob/main/electron/resources/telegram-instructions.md

Use this to acknowledge receipt of a task and inform the user you are starting to look into it.

```python
send_telegram("Got it! Looking into this...")
```

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

### Get memory stats

Source: https://github.com/charlie85270/dorothy/blob/main/skills/remember.md

Retrieves statistics about the memory system, such as the total number of memories stored or memory usage.

```APIDOC
## GET /api/memory/stats

### Description
Retrieves statistics about the memory system, such as the total number of memories stored or memory usage.

### Method
GET

### Endpoint
/api/memory/stats

### Parameters
(No parameters required)

### Request Example
```bash
curl -s http://127.0.0.1:31415/api/memory/stats
```

### Response
#### Success Response (200)
(No specific response schema provided in source, but typically returns memory statistics)

#### Response Example
(No specific response example provided in source)
```

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

### Configure SessionStart Hook

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/ai-dev-practice/prompts/15-slide-building-hooks.md

Defines a 'SessionStart' hook in settings.json, specifying the command to execute and a timeout in milliseconds.

```json
{
  "hooks": {
    "SessionStart": {
      "command": "~/.claude/hooks/start.sh",
      "timeout": 5000
    }
  }
}
```

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

### Get Sprite Catalog

Source: https://github.com/charlie85270/dorothy/blob/main/skills/world-builder/SKILL.md

Call this function once at the beginning of the automated world generation process to retrieve the available sprite catalog.

```lua
list_sprites
```

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

### Auto-Loading Project Context Hook

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/source-content.md

This bash script, intended for a SessionStart hook, uses a heredoc to output project details like name, stack, current sprint, key files, and recent Git changes.

```bash
# SessionStart hook
cat <<EOF
Project: E-commerce Platform
Stack: Node.js, React, PostgreSQL
Current Sprint: Authentication refactor
Key Files: src/auth/*, src/api/users.ts
Recent Changes: $(git log --oneline -3)
EOF
```

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

### Hook Configuration in settings.json

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/source-content.md

This JSON snippet shows how to configure various hooks, including SessionStart, UserPromptSubmit, and ToolCall (with before and after scripts), specifying their commands and timeouts.

```json
{
  "hooks": {
    "SessionStart": {
      "command": "~/.claude/hooks/session-start.sh",
      "timeout": 5000
    },
    "UserPromptSubmit": {
      "command": "~/.claude/hooks/inject-context.sh"
    },
    "ToolCall": {
      "before": "~/.claude/hooks/validate-action.sh",
      "after": "~/.claude/hooks/log-action.sh"
    }
  }
}
```

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

### mcp-orchestrator Agent Management Tools

Source: https://github.com/charlie85270/dorothy/blob/main/README.md

Tools for managing agents within the mcp-orchestrator server, including listing, creating, starting, stopping, and removing agents.

```APIDOC
## list_agents

### Description
List all agents with status, ID, name, project, and current task.

### Tool
`list_agents`

### Parameters
—

### Response
- Agent details including status, ID, name, project, and current task.
```

```APIDOC
## get_agent

### Description
Get detailed info about a specific agent including output history.

### Tool
`get_agent`

### Parameters
- **id** (string) - Required - The ID of the agent.

### Response
- Detailed agent information and output history.
```

```APIDOC
## get_agent_output

### Description
Read an agent's recent terminal output.

### Tool
`get_agent_output`

### Parameters
- **id** (string) - Required - The ID of the agent.
- **lines** (integer) - Optional - The number of lines to retrieve (default: 100).

### Response
- Recent terminal output of the agent.
```

```APIDOC
## create_agent

### Description
Create a new agent in idle state.

### Tool
`create_agent`

### Parameters
- **projectPath** (string) - Required - The project path for the agent.
- **name** (string) - Optional - The name of the agent.
- **skills** (string[]) - Optional - The skills the agent should possess.
- **character** (string) - Optional - The character description for the agent.
- **skipPermissions** (boolean) - Optional - Whether to skip permission checks (default: true).
- **secondaryProjectPath** (string) - Optional - A secondary project path for the agent.

### Response
- Information about the newly created agent.
```

```APIDOC
## start_agent

### Description
Start an agent with a task or send a message if already running.

### Tool
`start_agent`

### Parameters
- **id** (string) - Required - The ID of the agent.
- **prompt** (string) - Required - The task prompt for the agent.
- **model** (string) - Optional - The model to use for the agent.

### Response
- Status of the agent after starting or receiving the message.
```

```APIDOC
## send_message

### Description
Send input to a running agent (auto-starts idle agents).

### Tool
`send_message`

### Parameters
- **id** (string) - Required - The ID of the agent.
- **message** (string) - Required - The message to send to the agent.

### Response
- Status of the agent after receiving the message.
```

```APIDOC
## stop_agent

### Description
Terminate a running agent.

### Tool
`stop_agent`

### Parameters
- **id** (string) - Required - The ID of the agent.

### Response
- Status indicating the agent has been terminated.
```

```APIDOC
## remove_agent

### Description
Permanently delete an agent.

### Tool
`remove_agent`

### Parameters
- **id** (string) - Required - The ID of the agent.

### Response
- Confirmation of agent deletion.
```

```APIDOC
## wait_for_agent

### Description
Poll agent until completion, error, or waiting state.

### Tool
`wait_for_agent`

### Parameters
- **id** (string) - Required - The ID of the agent.
- **timeoutSeconds** (integer) - Optional - The maximum time to wait in seconds (default: 300).
- **pollIntervalSeconds** (integer) - Optional - The interval between polls in seconds (default: 5).

### Response
- Status indicating the agent's final state.
```

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

### Configure MCP Server (Production)

Source: https://github.com/charlie85270/dorothy/blob/main/mcp-orchestrator/README.md

Configuration for the MCP server in a production environment, specifying the command and arguments to run the orchestrator.

```json
{
  "mcpServers": {
    "claude-mgr-orchestrator": {
      "command": "node",
      "args": ["/path/to/dorothy/mcp-orchestrator/dist/index.js"]
    }
  }
}
```

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

### Basic Shell Scripting for Hooks

Source: https://github.com/charlie85270/dorothy/blob/main/slide-deck/source-content.md

This bash script demonstrates fundamental shell scripting practices for hooks, including making scripts executable, accessing environment variables, reading files, making external API calls with curl, and using exit codes for success or failure.

```bash
#!/bin/bash
# Make executable: chmod +x hook.sh

# Access environment variables
echo "User: $USER"
echo "PWD: $PWD"

# Read from files
cat .project-context.md

# Call external tools
curl -s api.example.com/status

# Exit codes
exit 0  # Success
exit 1  # Failure (blocks in some hooks)
```

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

### create_agent

Source: https://github.com/charlie85270/dorothy/blob/main/mcp-orchestrator/README.md

Creates a new agent for a specified project with optional configurations.

```APIDOC
## create_agent

### Description
Create a new agent for a specific project.

### Method
POST

### Endpoint
/agents

### Parameters
#### Request Body
- **projectPath** (string) - Required - Absolute path to the project directory.
- **name** (string) - Optional - Name for the agent.
- **skills** (array) - Optional - List of skill names to enable.
- **character** (string) - Optional - Visual character for the agent.
- **skipPermissions** (boolean) - Optional - Run with --dangerously-skip-permissions.
- **secondaryProjectPath** (string) - Optional - Additional project context.

### Response
#### Success Response (201)
- **agentId** (string) - The ID of the newly created agent.

#### Response Example
{
  "agentId": "agent-456"
}
```