@zerowidth/workbench-sdk 2.0.0-alpha.0

package/nodes/tool-cleanup/tool-cleanup.tests.json

@@ -0,0 +1,160 @@
+[
+  {
+    "description": "Passthrough when no tool calls present",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Hello"},
+        {"role": "assistant", "content": "Hi there!"},
+        {"role": "user", "content": "How are you?"}
+      ]
+    },
+    "expected": {
+      "messages": [
+        {"role": "user", "content": "Hello"},
+        {"role": "assistant", "content": "Hi there!"},
+        {"role": "user", "content": "How are you?"}
+      ],
+      "inserted_count": 0
+    }
+  },
+  {
+    "description": "Passthrough when all tool calls have responses",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "What's the weather?"},
+        {"role": "assistant", "content": null, "tool_calls": [{"id": "call_1", "function": {"name": "get_weather", "arguments": "{}"}}]},
+        {"role": "tool", "tool_call_id": "call_1", "content": "Sunny, 72F"},
+        {"role": "assistant", "content": "The weather is sunny and 72F!"}
+      ]
+    },
+    "expected": {
+      "messages": [
+        {"role": "user", "content": "What's the weather?"},
+        {"role": "assistant", "content": null, "tool_calls": [{"id": "call_1", "function": {"name": "get_weather", "arguments": "{}"}}]},
+        {"role": "tool", "tool_call_id": "call_1", "content": "Sunny, 72F"},
+        {"role": "assistant", "content": "The weather is sunny and 72F!"}
+      ],
+      "inserted_count": 0
+    }
+  },
+  {
+    "description": "Insert synthetic response when user message comes before tool response",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "What's the weather?"},
+        {"role": "assistant", "content": null, "tool_calls": [{"id": "call_1", "function": {"name": "get_weather", "arguments": "{}"}}]},
+        {"role": "user", "content": "Never mind"}
+      ]
+    },
+    "expected": {
+      "messages": [
+        {"role": "user", "content": "What's the weather?"},
+        {"role": "assistant", "content": null, "tool_calls": [{"id": "call_1", "function": {"name": "get_weather", "arguments": "{}"}}]},
+        {"role": "tool", "tool_call_id": "call_1", "content": "Tool call was not executed"},
+        {"role": "user", "content": "Never mind"}
+      ],
+      "inserted_count": 1
+    }
+  },
+  {
+    "description": "Insert multiple synthetic responses for multiple unanswered tool calls",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Get weather and time"},
+        {"role": "assistant", "content": null, "tool_calls": [
+          {"id": "call_1", "function": {"name": "get_weather", "arguments": "{}"}},
+          {"id": "call_2", "function": {"name": "get_time", "arguments": "{}"}}
+        ]},
+        {"role": "user", "content": "Cancel that"}
+      ]
+    },
+    "expected": {
+      "messages": [
+        {"role": "user", "content": "Get weather and time"},
+        {"role": "assistant", "content": null, "tool_calls": [
+          {"id": "call_1", "function": {"name": "get_weather", "arguments": "{}"}},
+          {"id": "call_2", "function": {"name": "get_time", "arguments": "{}"}}
+        ]},
+        {"role": "tool", "tool_call_id": "call_1", "content": "Tool call was not executed"},
+        {"role": "tool", "tool_call_id": "call_2", "content": "Tool call was not executed"},
+        {"role": "user", "content": "Cancel that"}
+      ],
+      "inserted_count": 2
+    }
+  },
+  {
+    "description": "Insert synthetic response at end of conversation",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "What's the weather?"},
+        {"role": "assistant", "content": null, "tool_calls": [{"id": "call_1", "function": {"name": "get_weather", "arguments": "{}"}}]}
+      ]
+    },
+    "expected": {
+      "messages": [
+        {"role": "user", "content": "What's the weather?"},
+        {"role": "assistant", "content": null, "tool_calls": [{"id": "call_1", "function": {"name": "get_weather", "arguments": "{}"}}]},
+        {"role": "tool", "tool_call_id": "call_1", "content": "Tool call was not executed"}
+      ],
+      "inserted_count": 1
+    }
+  },
+  {
+    "description": "Partial tool responses - backfill missing ones in order",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Get weather and time"},
+        {"role": "assistant", "content": null, "tool_calls": [
+          {"id": "call_1", "function": {"name": "get_weather", "arguments": "{}"}},
+          {"id": "call_2", "function": {"name": "get_time", "arguments": "{}"}},
+          {"id": "call_3", "function": {"name": "get_date", "arguments": "{}"}}
+        ]},
+        {"role": "tool", "tool_call_id": "call_2", "content": "3:00 PM"},
+        {"role": "user", "content": "Thanks"}
+      ]
+    },
+    "expected": {
+      "messages": [
+        {"role": "user", "content": "Get weather and time"},
+        {"role": "assistant", "content": null, "tool_calls": [
+          {"id": "call_1", "function": {"name": "get_weather", "arguments": "{}"}},
+          {"id": "call_2", "function": {"name": "get_time", "arguments": "{}"}},
+          {"id": "call_3", "function": {"name": "get_date", "arguments": "{}"}}
+        ]},
+        {"role": "tool", "tool_call_id": "call_1", "content": "Tool call was not executed"},
+        {"role": "tool", "tool_call_id": "call_2", "content": "3:00 PM"},
+        {"role": "tool", "tool_call_id": "call_3", "content": "Tool call was not executed"},
+        {"role": "user", "content": "Thanks"}
+      ],
+      "inserted_count": 2
+    }
+  },
+  {
+    "description": "Handle empty messages array",
+    "inputs": {
+      "messages": []
+    },
+    "expected": {
+      "messages": [],
+      "inserted_count": 0
+    }
+  },
+  {
+    "description": "Handle system messages passthrough",
+    "inputs": {
+      "messages": [
+        {"role": "system", "content": "You are a helpful assistant"},
+        {"role": "user", "content": "Hello"},
+        {"role": "assistant", "content": "Hi!"}
+      ]
+    },
+    "expected": {
+      "messages": [
+        {"role": "system", "content": "You are a helpful assistant"},
+        {"role": "user", "content": "Hello"},
+        {"role": "assistant", "content": "Hi!"}
+      ],
+      "inserted_count": 0
+    }
+  }
+]

package/nodes/truncate-by-tokens/truncate-by-tokens.config.json

@@ -0,0 +1,57 @@
+{
+  "display_name": "Truncate by Tokens",
+  "tagline": "Truncate by token limit",
+  "description": "Keep only the most recent messages that fit within the specified token limit.",
+  "icon": "scissors",
+  "category": "messages",
+  "inputs": [
+    {
+      "name": "messages",
+      "display_name": "Messages",
+      "type": "array",
+      "description": "Array of message objects to truncate",
+      "required": true
+    },
+    {
+      "name": "max_tokens",
+      "display_name": "Max Tokens",
+      "type": "number",
+      "description": "Maximum number of tokens to keep",
+      "required": true
+    }
+  ],
+  "outputs": [
+    {
+      "name": "messages",
+      "display_name": "Messages",
+      "type": "array",
+      "description": "Array containing messages that fit within the token limit"
+    },
+    {
+      "name": "token_count",
+      "display_name": "Token Count",
+      "type": "number",
+      "description": "Total number of tokens in the truncated messages"
+    },
+    {
+      "name": "truncated",
+      "display_name": "Truncated",
+      "type": "boolean",
+      "description": "Whether any messages were removed due to token limit"
+    }
+  ],
+  "settings": [
+    {
+      "name": "tokenizer",
+      "display_name": "Tokenizer",
+      "type": "string",
+      "description": "The tokenizer to use for counting tokens",
+      "default": "cl200k_base",
+      "options": [
+        "cl200k_base",
+        "cl100k_base"
+      ],
+      "required": true
+    }
+  ]
+}

package/nodes/truncate-by-tokens/truncate-by-tokens.process.js

@@ -0,0 +1,72 @@
+import { encoding_for_model } from 'tiktoken';
+
+export default async ({inputs, settings, config}) => {
+    try {
+        const messages = inputs.messages;
+        const maxTokens = inputs.max_tokens;
+
+        if (!Array.isArray(messages)) {
+            throw new Error("Messages input must be an array");
+        }
+
+        if (typeof maxTokens !== 'number' || maxTokens < 0 || !Number.isInteger(maxTokens)) {
+            throw new Error("Max tokens must be a non-negative integer");
+        }
+
+        const tokenizer = settings.tokenizer || 'cl200k_base';
+        if (!['cl200k_base', 'cl100k_base'].includes(tokenizer)) {
+            throw new Error(`Unsupported tokenizer: ${tokenizer}. Only cl200k_base and cl100k_base are supported.`);
+        }
+
+        const encoding = encoding_for_model(tokenizer === 'cl200k_base' ? 'gpt-4o' : 'gpt-4');
+
+        // Helper function to extract text content from message
+        const getMessageText = (message) => {
+            if (!message || typeof message !== 'object') return '';
+            if (typeof message.content === 'string') return message.content;
+            if (Array.isArray(message.content)) {
+                return message.content
+                    .filter(item => item && item.type === 'text' && item.text)
+                    .map(item => item.text)
+                    .join(' ');
+            }
+            return String(message.content || '');
+        };
+
+        // Helper function to count tokens in a message
+        const countMessageTokens = (message) => {
+            const text = getMessageText(message);
+            return encoding.encode(text).length;
+        };
+
+        // Start from the end and work backwards to find messages that fit
+        const result = [];
+        let totalTokens = 0;
+        let truncated = false;
+
+        for (let i = messages.length - 1; i >= 0; i--) {
+            const message = messages[i];
+            const messageTokens = countMessageTokens(message);
+            
+            if (totalTokens + messageTokens <= maxTokens) {
+                result.unshift(message); // Add to beginning to maintain order
+                totalTokens += messageTokens;
+            } else {
+                truncated = true;
+                break;
+            }
+        }
+
+        encoding.free();
+
+        return {
+            messages: result,
+            token_count: totalTokens,
+            truncated: truncated
+        };
+
+    } catch (error) {
+        console.log('Truncate by Tokens error:', error);
+        throw new Error(`Truncate by Tokens error: ${error.message}`);
+    }
+};

package/nodes/truncate-by-tokens/truncate-by-tokens.tests.json

@@ -0,0 +1,84 @@
+[
+  {
+    "description": "Truncate messages to fit within token limit",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Hello there, how are you doing today? I hope you're having a wonderful day."},
+        {"role": "assistant", "content": "Hi there! I'm doing great, thank you for asking. I'm here to help you with any questions you might have."},
+        {"role": "user", "content": "That's fantastic to hear! I really appreciate your help and support."},
+        {"role": "assistant", "content": "You're very welcome! I'm always happy to assist you with whatever you need. Feel free to ask me anything."}
+      ],
+      "max_tokens": 20
+    },
+    "expected": {
+      "truncated": true
+    }
+  },
+  {
+    "description": "All messages fit within token limit",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Hi"},
+        {"role": "assistant", "content": "Hello!"}
+      ],
+      "max_tokens": 50
+    },
+    "expected": {
+      "truncated": false
+    }
+  },
+  {
+    "description": "Empty message array",
+    "inputs": {
+      "messages": [],
+      "max_tokens": 10
+    },
+    "expected": {
+      "token_count": 0,
+      "truncated": false
+    }
+  },
+  {
+    "description": "Zero token limit",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Hello"},
+        {"role": "assistant", "content": "Hi there!"}
+      ],
+      "max_tokens": 0
+    },
+    "expected": {
+      "token_count": 0,
+      "truncated": true
+    }
+  },
+  {
+    "description": "Single message that exceeds limit",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "This is a very long message that will exceed the token limit"}
+      ],
+      "max_tokens": 5
+    },
+    "expected": {
+      "token_count": 0,
+      "truncated": true
+    }
+  },
+  {
+    "description": "Multi-modal content handling",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Hello"},
+        {"role": "assistant", "content": [
+          {"type": "text", "text": "Hi there!"},
+          {"type": "image_url", "image_url": {"url": "https://example.com/image.png"}}
+        ]}
+      ],
+      "max_tokens": 10
+    },
+    "expected": {
+      "truncated": false
+    }
+  }
+]

package/nodes/truncate-by-tokens-from-start/truncate-by-tokens-from-start.config.json

@@ -0,0 +1,57 @@
+{
+  "display_name": "Truncate by Tokens from Start",
+  "tagline": "Truncate from start",
+  "description": "Keep only the first messages that fit within the specified token limit.",
+  "icon": "scissors",
+  "category": "messages",
+  "inputs": [
+    {
+      "name": "messages",
+      "display_name": "Messages",
+      "type": "array",
+      "description": "Array of message objects to truncate",
+      "required": true
+    },
+    {
+      "name": "max_tokens",
+      "display_name": "Max Tokens",
+      "type": "number",
+      "description": "Maximum number of tokens to keep",
+      "required": true
+    }
+  ],
+  "outputs": [
+    {
+      "name": "messages",
+      "display_name": "Messages",
+      "type": "array",
+      "description": "Array containing messages that fit within the token limit"
+    },
+    {
+      "name": "token_count",
+      "display_name": "Token Count",
+      "type": "number",
+      "description": "Total number of tokens in the truncated messages"
+    },
+    {
+      "name": "truncated",
+      "display_name": "Truncated",
+      "type": "boolean",
+      "description": "Whether any messages were removed due to token limit"
+    }
+  ],
+  "settings": [
+    {
+      "name": "tokenizer",
+      "display_name": "Tokenizer",
+      "type": "string",
+      "description": "The tokenizer to use for counting tokens",
+      "default": "cl200k_base",
+      "options": [
+        "cl200k_base",
+        "cl100k_base"
+      ],
+      "required": true
+    }
+  ]
+}

package/nodes/truncate-by-tokens-from-start/truncate-by-tokens-from-start.process.js

@@ -0,0 +1,72 @@
+import { encoding_for_model } from 'tiktoken';
+
+export default async ({inputs, settings, config}) => {
+    try {
+        const messages = inputs.messages;
+        const maxTokens = inputs.max_tokens;
+
+        if (!Array.isArray(messages)) {
+            throw new Error("Messages input must be an array");
+        }
+
+        if (typeof maxTokens !== 'number' || maxTokens < 0 || !Number.isInteger(maxTokens)) {
+            throw new Error("Max tokens must be a non-negative integer");
+        }
+
+        const tokenizer = settings.tokenizer || 'cl200k_base';
+        if (!['cl200k_base', 'cl100k_base'].includes(tokenizer)) {
+            throw new Error(`Unsupported tokenizer: ${tokenizer}. Only cl200k_base and cl100k_base are supported.`);
+        }
+
+        const encoding = encoding_for_model(tokenizer === 'cl200k_base' ? 'gpt-4o' : 'gpt-4');
+
+        // Helper function to extract text content from message
+        const getMessageText = (message) => {
+            if (!message || typeof message !== 'object') return '';
+            if (typeof message.content === 'string') return message.content;
+            if (Array.isArray(message.content)) {
+                return message.content
+                    .filter(item => item && item.type === 'text' && item.text)
+                    .map(item => item.text)
+                    .join(' ');
+            }
+            return String(message.content || '');
+        };
+
+        // Helper function to count tokens in a message
+        const countMessageTokens = (message) => {
+            const text = getMessageText(message);
+            return encoding.encode(text).length;
+        };
+
+        // Start from the beginning and work forwards to find messages that fit
+        const result = [];
+        let totalTokens = 0;
+        let truncated = false;
+
+        for (let i = 0; i < messages.length; i++) {
+            const message = messages[i];
+            const messageTokens = countMessageTokens(message);
+            
+            if (totalTokens + messageTokens <= maxTokens) {
+                result.push(message);
+                totalTokens += messageTokens;
+            } else {
+                truncated = true;
+                break;
+            }
+        }
+
+        encoding.free();
+
+        return {
+            messages: result,
+            token_count: totalTokens,
+            truncated: truncated
+        };
+
+    } catch (error) {
+        console.log('Truncate by Tokens from Start error:', error);
+        throw new Error(`Truncate by Tokens from Start error: ${error.message}`);
+    }
+};

package/nodes/truncate-by-tokens-from-start/truncate-by-tokens-from-start.tests.json

@@ -0,0 +1,84 @@
+[
+  {
+    "description": "Truncate messages from start to fit within token limit",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Hello there, how are you doing today? I hope you're having a wonderful day."},
+        {"role": "assistant", "content": "Hi there! I'm doing great, thank you for asking. I'm here to help you with any questions you might have."},
+        {"role": "user", "content": "That's fantastic to hear! I really appreciate your help and support."},
+        {"role": "assistant", "content": "You're very welcome! I'm always happy to assist you with whatever you need. Feel free to ask me anything."}
+      ],
+      "max_tokens": 20
+    },
+    "expected": {
+      "truncated": true
+    }
+  },
+  {
+    "description": "All messages fit within token limit",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Hi"},
+        {"role": "assistant", "content": "Hello!"}
+      ],
+      "max_tokens": 50
+    },
+    "expected": {
+      "truncated": false
+    }
+  },
+  {
+    "description": "Empty message array",
+    "inputs": {
+      "messages": [],
+      "max_tokens": 10
+    },
+    "expected": {
+      "token_count": 0,
+      "truncated": false
+    }
+  },
+  {
+    "description": "Zero token limit",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Hello"},
+        {"role": "assistant", "content": "Hi there!"}
+      ],
+      "max_tokens": 0
+    },
+    "expected": {
+      "token_count": 0,
+      "truncated": true
+    }
+  },
+  {
+    "description": "Single message that exceeds limit",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "This is a very long message that will exceed the token limit"}
+      ],
+      "max_tokens": 5
+    },
+    "expected": {
+      "token_count": 0,
+      "truncated": true
+    }
+  },
+  {
+    "description": "Multi-modal content handling",
+    "inputs": {
+      "messages": [
+        {"role": "user", "content": "Hello"},
+        {"role": "assistant", "content": [
+          {"type": "text", "text": "Hi there!"},
+          {"type": "image_url", "image_url": {"url": "https://example.com/image.png"}}
+        ]}
+      ],
+      "max_tokens": 10
+    },
+    "expected": {
+      "truncated": false
+    }
+  }
+]

package/nodes/truncate-by-tokens-preserve-system/truncate-by-tokens-preserve-system.config.json

@@ -0,0 +1,71 @@
+{
+  "display_name": "Truncate by Tokens Preserve System",
+  "tagline": "Truncate preserving system",
+  "description": "Always keep system messages, truncate other messages to fit within the specified token limit.",
+  "icon": "shield",
+  "category": "messages",
+  "inputs": [
+    {
+      "name": "messages",
+      "display_name": "Messages",
+      "type": "array",
+      "description": "Array of message objects to truncate",
+      "required": true
+    },
+    {
+      "name": "max_tokens",
+      "display_name": "Max Tokens",
+      "type": "number",
+      "description": "Maximum number of tokens to keep",
+      "required": true
+    }
+  ],
+  "outputs": [
+    {
+      "name": "messages",
+      "display_name": "Messages",
+      "type": "array",
+      "description": "Array containing system messages and other messages that fit within the token limit"
+    },
+    {
+      "name": "token_count",
+      "display_name": "Token Count",
+      "type": "number",
+      "description": "Total number of tokens in the truncated messages"
+    },
+    {
+      "name": "truncated",
+      "display_name": "Truncated",
+      "type": "boolean",
+      "description": "Whether any non-system messages were removed due to token limit"
+    },
+    {
+      "name": "system_count",
+      "display_name": "System Count",
+      "type": "number",
+      "description": "Number of system messages preserved"
+    }
+  ],
+  "settings": [
+    {
+      "name": "tokenizer",
+      "display_name": "Tokenizer",
+      "type": "string",
+      "description": "The tokenizer to use for counting tokens",
+      "default": "cl200k_base",
+      "options": [
+        "cl200k_base",
+        "cl100k_base"
+      ],
+      "required": true
+    },
+    {
+      "name": "count_system",
+      "display_name": "Count System",
+      "type": "boolean",
+      "description": "Whether system messages count toward the token limit (false = always include system messages)",
+      "default": false,
+      "required": true
+    }
+  ]
+}