Anthropic support

.github/workflows/e2e_tests.yml

@@ -2,8 +2,6 @@
 name: e2e tests
 on:
   pull_request:
-    branches:
-      - main
 jobs:
   e2e-tests:
     runs-on: ubuntu-latest
@@ -71,6 +69,7 @@ jobs:
         E2B_API_KEY: ${{ secrets.E2B_API_KEY }}
         PINATA_GATEWAY_TOKEN: ${{ secrets.PINATA_GATEWAY_TOKEN }}
         PINATA_API_JWT: ${{ secrets.PINATA_API_JWT }}
+        ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
     - name: "OpenAI gpt-4-turbo-preview"
       run:
         cd contracts && npx hardhat openai --contract-address ${{ env.TEST_CONTRACT_ADDRESS }} --model gpt-4-turbo-preview --message "Who is the president of USA?" --network ${{ env.NETWORK }}
@@ -111,6 +110,11 @@ jobs:
         cd contracts && npx hardhat groq --contract-address ${{ env.TEST_CONTRACT_ADDRESS }} --model gemma-7b-it --message "Who is the president of USA?" --network ${{ env.NETWORK }}
       env:
         PRIVATE_KEY_LOCALHOST: ${{ secrets.PRIVATE_KEY }}
+    - name: "Anthropic claude-3-5-sonnet-20240620"
+      run:
+        cd contracts && npx hardhat llm --contract-address ${{ env.TEST_CONTRACT_ADDRESS }} --model claude-3-5-sonnet-20240620 --message "Who is the president of USA?" --network ${{ env.NETWORK }}
+      env:
+        PRIVATE_KEY_LOCALHOST: ${{ secrets.PRIVATE_KEY }}
     - name: "OpenAI Image Generation"
       run:
         cd contracts && npx hardhat image_generation --contract-address ${{ env.TEST_CONTRACT_ADDRESS }} --query "Red rose" --network ${{ env.NETWORK }}

contracts/contracts/ChatGpt.sol

@@ -32,10 +32,16 @@ contract ChatGpt {
 
     // @notice Address of the oracle contract
     address public oracleAddress;
+
+    // @notice Configuration for the LLM request
+    IOracle.LlmRequest private config;
 
     // @notice CID of the knowledge base
     string public knowledgeBase;
 
+    // @notice Mapping from chat ID to the tool currently running
+    mapping(uint => string) public toolRunning;
+
     // @notice Event emitted when the oracle address is updated
     event OracleAddressUpdated(address indexed newOracleAddress);
 
@@ -45,6 +51,22 @@ contract ChatGpt {
         owner = msg.sender;
         oracleAddress = initialOracleAddress;
         knowledgeBase = knowledgeBaseCID;
+
+        config = IOracle.LlmRequest({
+            model : "claude-3-5-sonnet-20240620",
+            frequencyPenalty : 21, // > 20 for null
+            logitBias : "", // empty str for null
+            maxTokens : 1000, // 0 for null
+            presencePenalty : 21, // > 20 for null
+            responseFormat : "{\"type\":\"text\"}",
+            seed : 0, // null
+            stop : "", // null
+            temperature : 10, // Example temperature (scaled up, 10 means 1.0), > 20 means null
+            topP : 101, // Percentage 0-100, > 100 means null
+            tools : "[{\"type\":\"function\",\"function\":{\"name\":\"web_search\",\"description\":\"Search the internet\",\"parameters\":{\"type\":\"object\",\"properties\":{\"query\":{\"type\":\"string\",\"description\":\"Search query\"}},\"required\":[\"query\"]}}},{\"type\":\"function\",\"function\":{\"name\":\"code_interpreter\",\"description\":\"Evaluates python code in a sandbox environment. The environment resets on every execution. You must send the whole script every time and print your outputs. Script should be pure python code that can be evaluated. It should be in python format NOT markdown. The code should NOT be wrapped in backticks. All python packages including requests, matplotlib, scipy, numpy, pandas, etc are available. Output can only be read from stdout, and stdin. Do not use things like plot.show() as it will not work. print() any output and results so you can capture the output.\",\"parameters\":{\"type\":\"object\",\"properties\":{\"code\":{\"type\":\"string\",\"description\":\"The pure python script to be evaluated. The contents will be in main.py. It should not be in markdown format.\"}},\"required\":[\"code\"]}}}]",
+            toolChoice : "auto", // "none" or "auto"
+            user : "" // null
+        });
     }
 
     // @notice Ensures the caller is the contract owner
@@ -92,7 +114,7 @@ contract ChatGpt {
             );
         } else {
             // Otherwise, create an LLM call
-            IOracle(oracleAddress).createLlmCall(currentId);
+            IOracle(oracleAddress).createLlmCall(currentId, config);
         }
         emit ChatCreated(msg.sender, currentId);
 
@@ -105,8 +127,8 @@ contract ChatGpt {
     // @dev Called by teeML oracle
     function onOracleLlmResponse(
         uint runId,
-        string memory response,
-        string memory /*errorMessage*/
+        IOracle.LlmResponse memory response,
+        string memory errorMessage
     ) public onlyOracle {
         ChatRun storage run = chatRuns[runId];
         require(
@@ -115,10 +137,48 @@ contract ChatGpt {
         );
 
         Message memory newMessage;
-        newMessage.content = response;
-        newMessage.role = "assistant";
-        run.messages.push(newMessage);
-        run.messagesCount++;
+        if (!compareStrings(errorMessage, "")) {
+            newMessage.role = "assistant";
+            newMessage.content = errorMessage;
+            run.messages.push(newMessage);
+            run.messagesCount++;
+        } else {
+            if (!compareStrings(response.functionName, "")) {
+                toolRunning[runId] = response.functionName;
+                IOracle(oracleAddress).createFunctionCall(runId, response.functionName, response.functionArguments);
+            } else {
+                toolRunning[runId] = "";
+            }
+            newMessage.role = "assistant";
+            newMessage.content = response.content;
+            run.messages.push(newMessage);
+            run.messagesCount++;
+        }
+    }
+
+    // @notice Handles the response from the oracle for a function call
+    // @param runId The ID of the chat run
+    // @param response The response from the oracle
+    // @param errorMessage Any error message
+    // @dev Called by teeML oracle
+    function onOracleFunctionResponse(
+        uint runId,
+        string memory response,
+        string memory errorMessage
+    ) public onlyOracle {
+        require(
+            !compareStrings(toolRunning[runId], ""),
+            "No function to respond to"
+        );
+        ChatRun storage run = chatRuns[runId];
+        if (compareStrings(errorMessage, "")) {
+            Message memory newMessage;
+            newMessage.role = "user";
+            newMessage.content = response;
+            run.messages.push(newMessage);
+            run.messagesCount++;
+            IOracle(oracleAddress).createLlmCall(runId, config);
+        }
     }
 
     // @notice Handles the response from the oracle for a knowledge base query
@@ -155,7 +215,7 @@ contract ChatGpt {
         lastMessage.content = newContent;
 
         // Call LLM
-        IOracle(oracleAddress).createLlmCall(runId);
+        IOracle(oracleAddress).createLlmCall(runId, config);
     }
 
     // @notice Adds a new message to an existing chat run
@@ -186,7 +246,7 @@ contract ChatGpt {
             );
         } else {
             // Otherwise, create an LLM call
-            IOracle(oracleAddress).createLlmCall(runId);
+            IOracle(oracleAddress).createLlmCall(runId, config);
         }
     }
 
@@ -213,4 +273,12 @@ contract ChatGpt {
         }
         return roles;
     }
+
+    // @notice Compares two strings for equality
+    // @param a The first string
+    // @param b The second string
+    // @return True if the strings are equal, false otherwise
+    function compareStrings(string memory a, string memory b) private pure returns (bool) {
+        return (keccak256(abi.encodePacked((a))) == keccak256(abi.encodePacked((b))));
+    }
 }

contracts/contracts/ChatOracle.sol

@@ -43,6 +43,9 @@ contract ChatOracle is IOracle {
     // @dev Default is OpenAI
     mapping(uint => string) public promptType;
 
+    // @notice Mapping of prompt ID to the LLM configuration
+    mapping(uint => IOracle.LlmRequest) public llmConfigurations;
+
     // @notice Mapping of prompt ID to the OpenAI configuration
     mapping(uint => IOracle.OpenAiRequest) public openAiConfigurations;
 
@@ -178,7 +181,7 @@ contract ChatOracle is IOracle {
     // @notice Creates a new LLM call
     // @param promptCallbackId The callback ID for the LLM call
     // @return The ID of the created prompt
-    function createLlmCall(uint promptCallbackId) public returns (uint) {
+    function createLlmCall(uint promptCallbackId, IOracle.LlmRequest memory request) public returns (uint) {
         uint promptId = promptsCount;
         callbackAddresses[promptId] = msg.sender;
         promptCallbackIds[promptId] = promptCallbackId;
@@ -187,6 +190,7 @@ contract ChatOracle is IOracle {
 
         promptsCount++;
 
+        llmConfigurations[promptId] = request;
         emit PromptAdded(promptId, promptCallbackId, msg.sender);
 
         return promptId;
@@ -201,7 +205,7 @@ contract ChatOracle is IOracle {
     function addResponse(
         uint promptId,
         uint promptCallBackId,
-        string memory response,
+        IOracle.LlmResponse memory response,
         string memory errorMessage
     ) public onlyWhitelisted {
         require(!isPromptProcessed[promptId], "Prompt already processed");

contracts/contracts/GroqChatGpt.sol

@@ -33,6 +33,9 @@ contract GroqChatGpt {
     // @notice Address of the oracle contract
     address public oracleAddress;
 
+    // @notice Mapping from chat ID to the tool currently running
+    mapping(uint => string) public toolRunning;
+
     // @notice Event emitted when the oracle address is updated
     event OracleAddressUpdated(address indexed newOracleAddress);
 
@@ -56,6 +59,8 @@ contract GroqChatGpt {
         stop : "", // null
         temperature : 10, // Example temperature (scaled up, 10 means 1.0), > 20 means null
         topP : 101, // Percentage 0-100, > 100 means null
+        tools : "[{\"type\":\"function\",\"function\":{\"name\":\"web_search\",\"description\":\"Search the internet\",\"parameters\":{\"type\":\"object\",\"properties\":{\"query\":{\"type\":\"string\",\"description\":\"Search query\"}},\"required\":[\"query\"]}}},{\"type\":\"function\",\"function\":{\"name\":\"code_interpreter\",\"description\":\"Evaluates python code in a sandbox environment. The environment resets on every execution. You must send the whole script every time and print your outputs. Script should be pure python code that can be evaluated. It should be in python format NOT markdown. The code should NOT be wrapped in backticks. All python packages including requests, matplotlib, scipy, numpy, pandas, etc are available. Output can only be read from stdout, and stdin. Do not use things like plot.show() as it will not work. print() any output and results so you can capture the output.\",\"parameters\":{\"type\":\"object\",\"properties\":{\"code\":{\"type\":\"string\",\"description\":\"The pure python script to be evaluated. The contents will be in main.py. It should not be in markdown format.\"}},\"required\":[\"code\"]}}}]",
+        toolChoice : "auto", // "none" or "auto"
         user : "" // null
         });
     }
@@ -122,6 +127,12 @@ contract GroqChatGpt {
             run.messages.push(newMessage);
             run.messagesCount++;
         } else {
+            if (!compareStrings(response.functionName, "")) {
+                toolRunning[runId] = response.functionName;
+                IOracle(oracleAddress).createFunctionCall(runId, response.functionName, response.functionArguments);
+            } else {
+                toolRunning[runId] = "";
+            }
             Message memory newMessage;
             newMessage.role = "assistant";
             newMessage.content = response.content;
@@ -130,6 +141,31 @@ contract GroqChatGpt {
         }
     }
 
+    // @notice Handles the response from the oracle for a function call
+    // @param runId The ID of the chat run
+    // @param response The response from the oracle
+    // @param errorMessage Any error message
+    // @dev Called by teeML oracle
+    function onOracleFunctionResponse(
+        uint runId,
+        string memory response,
+        string memory errorMessage
+    ) public onlyOracle {
+        require(
+            !compareStrings(toolRunning[runId], ""),
+            "No function to respond to"
+        );
+        ChatRun storage run = chatRuns[runId];
+        if (compareStrings(errorMessage, "")) {
+            Message memory newMessage;
+            newMessage.role = "user";
+            newMessage.content = response;
+            run.messages.push(newMessage);
+            run.messagesCount++;
+            IOracle(oracleAddress).createGroqLlmCall(runId, config);
+        }
+    }
+
     // @notice Adds a new message to an existing chat run
     // @param message The new message to add
     // @param runId The ID of the chat run

contracts/contracts/OpenAiChatGpt.sol

@@ -33,6 +33,9 @@ contract OpenAiChatGpt {
     // @notice Address of the oracle contract
     address public oracleAddress;
 
+    // @notice Mapping from chat ID to the tool currently running
+    mapping(uint => string) public toolRunning;
+
     // @notice Event emitted when the oracle address is updated
     event OracleAddressUpdated(address indexed newOracleAddress);
 
@@ -46,19 +49,19 @@ contract OpenAiChatGpt {
         chatRunsCount = 0;
 
         config = IOracle.OpenAiRequest({
-        model : "gpt-4-turbo-preview",
-        frequencyPenalty : 21, // > 20 for null
-        logitBias : "", // empty str for null
-        maxTokens : 1000, // 0 for null
-        presencePenalty : 21, // > 20 for null
-        responseFormat : "{\"type\":\"text\"}",
-        seed : 0, // null
-        stop : "", // null
-        temperature : 10, // Example temperature (scaled up, 10 means 1.0), > 20 means null
-        topP : 101, // Percentage 0-100, > 100 means null
-        tools : "[{\"type\":\"function\",\"function\":{\"name\":\"web_search\",\"description\":\"Search the internet\",\"parameters\":{\"type\":\"object\",\"properties\":{\"query\":{\"type\":\"string\",\"description\":\"Search query\"}},\"required\":[\"query\"]}}},{\"type\":\"function\",\"function\":{\"name\":\"code_interpreter\",\"description\":\"Evaluates python code in a sandbox environment. The environment resets on every execution. You must send the whole script every time and print your outputs. Script should be pure python code that can be evaluated. It should be in python format NOT markdown. The code should NOT be wrapped in backticks. All python packages including requests, matplotlib, scipy, numpy, pandas, etc are available. Output can only be read from stdout, and stdin. Do not use things like plot.show() as it will not work. print() any output and results so you can capture the output.\",\"parameters\":{\"type\":\"object\",\"properties\":{\"code\":{\"type\":\"string\",\"description\":\"The pure python script to be evaluated. The contents will be in main.py. It should not be in markdown format.\"}},\"required\":[\"code\"]}}}]",
-        toolChoice : "auto", // "none" or "auto"
-        user : "" // null
+            model : "gpt-4-turbo-preview",
+            frequencyPenalty : 21, // > 20 for null
+            logitBias : "", // empty str for null
+            maxTokens : 1000, // 0 for null
+            presencePenalty : 21, // > 20 for null
+            responseFormat : "{\"type\":\"text\"}",
+            seed : 0, // null
+            stop : "", // null
+            temperature : 10, // Example temperature (scaled up, 10 means 1.0), > 20 means null
+            topP : 101, // Percentage 0-100, > 100 means null
+            tools : "[{\"type\":\"function\",\"function\":{\"name\":\"web_search\",\"description\":\"Search the internet\",\"parameters\":{\"type\":\"object\",\"properties\":{\"query\":{\"type\":\"string\",\"description\":\"Search query\"}},\"required\":[\"query\"]}}},{\"type\":\"function\",\"function\":{\"name\":\"code_interpreter\",\"description\":\"Evaluates python code in a sandbox environment. The environment resets on every execution. You must send the whole script every time and print your outputs. Script should be pure python code that can be evaluated. It should be in python format NOT markdown. The code should NOT be wrapped in backticks. All python packages including requests, matplotlib, scipy, numpy, pandas, etc are available. Output can only be read from stdout, and stdin. Do not use things like plot.show() as it will not work. print() any output and results so you can capture the output.\",\"parameters\":{\"type\":\"object\",\"properties\":{\"code\":{\"type\":\"string\",\"description\":\"The pure python script to be evaluated. The contents will be in main.py. It should not be in markdown format.\"}},\"required\":[\"code\"]}}}]",
+            toolChoice : "auto", // "none" or "auto"
+            user : "" // null
         });
     }
 
@@ -126,15 +129,17 @@ contract OpenAiChatGpt {
             run.messages.push(newMessage);
             run.messagesCount++;
         } else {
-            if (compareStrings(response.content, "")) {
+            if (!compareStrings(response.functionName, "")) {
+                toolRunning[runId] = response.functionName;
                 IOracle(oracleAddress).createFunctionCall(runId, response.functionName, response.functionArguments);
             } else {
-                Message memory newMessage;
-                newMessage.role = "assistant";
-                newMessage.content = response.content;
-                run.messages.push(newMessage);
-                run.messagesCount++;
+                toolRunning[runId] = "";
             }
+            Message memory newMessage;
+            newMessage.role = "assistant";
+            newMessage.content = response.content;
+            run.messages.push(newMessage);
+            run.messagesCount++;
         }
     }
 
@@ -148,11 +153,11 @@ contract OpenAiChatGpt {
         string memory response,
         string memory errorMessage
     ) public onlyOracle {
-        ChatRun storage run = chatRuns[runId];
         require(
-            compareStrings(run.messages[run.messagesCount - 1].role, "user"),
+            !compareStrings(toolRunning[runId], ""),
             "No function to respond to"
         );
+        ChatRun storage run = chatRuns[runId];
         if (compareStrings(errorMessage, "")) {
             Message memory newMessage;
             newMessage.role = "user";