@langchain/classic 1.0.39 → 1.0.41

package/dist/evaluation/qa/eval_chain.js

@@ -8,12 +8,13 @@ var QAEvalChain = class extends LLMChain {
 	}
 	static fromLlm(llm, options = {}) {
 		const prompt = options.prompt || QA_PROMPT;
-		const expectedInputVars = new Set([
+		const expectedInputVars = /* @__PURE__ */ new Set([
 			"query",
 			"answer",
 			"result"
 		]);
-		if (!eqSet(expectedInputVars, new Set(prompt.inputVariables))) throw new Error(`Input variables should be ${[...expectedInputVars]}, but got ${prompt.inputVariables}`);
+		const inputVarsSet = new Set(prompt.inputVariables);
+		if (!eqSet(expectedInputVars, inputVarsSet)) throw new Error(`Input variables should be ${[...expectedInputVars]}, but got ${prompt.inputVariables}`);
 		return new this({
 			llm,
 			prompt,

package/dist/evaluation/qa/eval_chain.js.map

@@ -1 +1 @@
-{"version":3,"file":"eval_chain.js","names":[],"sources":["../../../src/evaluation/qa/eval_chain.ts"],"sourcesContent":["import type { BaseLanguageModelInterface } from \"@langchain/core/language_models/base\";\nimport { ChainValues } from \"@langchain/core/utils/types\";\nimport { PromptTemplate } from \"@langchain/core/prompts\";\nimport { QA_PROMPT } from \"./prompt.js\";\nimport { LLMChain, LLMChainInput } from \"../../chains/llm_chain.js\";\n\nexport interface EvaluateArgs {\n  questionKey: string;\n  answerKey: string;\n  predictionKey: string;\n}\n\nconst eqSet = (xs: Set<string>, ys: Set<string>) =>\n  xs.size === ys.size && [...xs].every((x) => ys.has(x));\n\nexport class QAEvalChain extends LLMChain {\n  static lc_name() {\n    return \"QAEvalChain\";\n  }\n\n  static fromLlm(\n    llm: BaseLanguageModelInterface,\n    options: {\n      prompt?: PromptTemplate;\n      chainInput?: Omit<LLMChainInput, \"llm\">;\n    } = {}\n  ): QAEvalChain {\n    const prompt = options.prompt || QA_PROMPT;\n    const expectedInputVars: Set<string> = new Set([\n      \"query\",\n      \"answer\",\n      \"result\",\n    ]);\n    // Create a Set from inputVariables for a valid comparison\n    const inputVarsSet: Set<string> = new Set(prompt.inputVariables);\n\n    if (!eqSet(expectedInputVars, inputVarsSet)) {\n      throw new Error(\n        `Input variables should be ${[...expectedInputVars]}, but got ${\n          prompt.inputVariables\n        }`\n      );\n    }\n    return new this({ llm, prompt, ...options.chainInput });\n  }\n\n  async evaluate(\n    examples: ChainValues,\n    predictions: ChainValues,\n    args: EvaluateArgs = {\n      questionKey: \"query\",\n      answerKey: \"answer\",\n      predictionKey: \"result\",\n    }\n  ): Promise<ChainValues> {\n    const inputs = examples.map((example: ChainValues, i: number) => ({\n      query: example[args.questionKey],\n      answer: example[args.answerKey],\n      result: predictions[i][args.predictionKey],\n    }));\n\n    return await this.apply(inputs);\n  }\n}\n"],"mappings":";;;AAYA,MAAM,SAAS,IAAiB,OAC9B,GAAG,SAAS,GAAG,QAAQ,CAAC,GAAG,GAAG,CAAC,OAAO,MAAM,GAAG,IAAI,EAAE,CAAC;AAExD,IAAa,cAAb,cAAiC,SAAS;CACxC,OAAO,UAAU;AACf,SAAO;;CAGT,OAAO,QACL,KACA,UAGI,EAAE,EACO;EACb,MAAM,SAAS,QAAQ,UAAU;EACjC,MAAM,oBAAiC,IAAI,IAAI;GAC7C;GACA;GACA;GACD,CAAC;AAIF,MAAI,CAAC,MAAM,mBAAmB,IAFQ,IAAI,OAAO,eAEP,CAAC,CACzC,OAAM,IAAI,MACR,6BAA6B,CAAC,GAAG,kBAAkB,CAAC,YAClD,OAAO,iBAEV;AAEH,SAAO,IAAI,KAAK;GAAE;GAAK;GAAQ,GAAG,QAAQ;GAAY,CAAC;;CAGzD,MAAM,SACJ,UACA,aACA,OAAqB;EACnB,aAAa;EACb,WAAW;EACX,eAAe;EAChB,EACqB;EACtB,MAAM,SAAS,SAAS,KAAK,SAAsB,OAAe;GAChE,OAAO,QAAQ,KAAK;GACpB,QAAQ,QAAQ,KAAK;GACrB,QAAQ,YAAY,GAAG,KAAK;GAC7B,EAAE;AAEH,SAAO,MAAM,KAAK,MAAM,OAAO"}
+{"version":3,"file":"eval_chain.js","names":[],"sources":["../../../src/evaluation/qa/eval_chain.ts"],"sourcesContent":["import type { BaseLanguageModelInterface } from \"@langchain/core/language_models/base\";\nimport { ChainValues } from \"@langchain/core/utils/types\";\nimport { PromptTemplate } from \"@langchain/core/prompts\";\nimport { QA_PROMPT } from \"./prompt.js\";\nimport { LLMChain, LLMChainInput } from \"../../chains/llm_chain.js\";\n\nexport interface EvaluateArgs {\n  questionKey: string;\n  answerKey: string;\n  predictionKey: string;\n}\n\nconst eqSet = (xs: Set<string>, ys: Set<string>) =>\n  xs.size === ys.size && [...xs].every((x) => ys.has(x));\n\nexport class QAEvalChain extends LLMChain {\n  static lc_name() {\n    return \"QAEvalChain\";\n  }\n\n  static fromLlm(\n    llm: BaseLanguageModelInterface,\n    options: {\n      prompt?: PromptTemplate;\n      chainInput?: Omit<LLMChainInput, \"llm\">;\n    } = {}\n  ): QAEvalChain {\n    const prompt = options.prompt || QA_PROMPT;\n    const expectedInputVars: Set<string> = new Set([\n      \"query\",\n      \"answer\",\n      \"result\",\n    ]);\n    // Create a Set from inputVariables for a valid comparison\n    const inputVarsSet: Set<string> = new Set(prompt.inputVariables);\n\n    if (!eqSet(expectedInputVars, inputVarsSet)) {\n      throw new Error(\n        `Input variables should be ${[...expectedInputVars]}, but got ${\n          prompt.inputVariables\n        }`\n      );\n    }\n    return new this({ llm, prompt, ...options.chainInput });\n  }\n\n  async evaluate(\n    examples: ChainValues,\n    predictions: ChainValues,\n    args: EvaluateArgs = {\n      questionKey: \"query\",\n      answerKey: \"answer\",\n      predictionKey: \"result\",\n    }\n  ): Promise<ChainValues> {\n    const inputs = examples.map((example: ChainValues, i: number) => ({\n      query: example[args.questionKey],\n      answer: example[args.answerKey],\n      result: predictions[i][args.predictionKey],\n    }));\n\n    return await this.apply(inputs);\n  }\n}\n"],"mappings":";;;AAYA,MAAM,SAAS,IAAiB,OAC9B,GAAG,SAAS,GAAG,QAAQ,CAAC,GAAG,EAAE,CAAC,CAAC,OAAO,MAAM,GAAG,IAAI,CAAC,CAAC;AAEvD,IAAa,cAAb,cAAiC,SAAS;CACxC,OAAO,UAAU;EACf,OAAO;CACT;CAEA,OAAO,QACL,KACA,UAGI,CAAC,GACQ;EACb,MAAM,SAAS,QAAQ,UAAU;EACjC,MAAM,oCAAiC,IAAI,IAAI;GAC7C;GACA;GACA;EACF,CAAC;EAED,MAAM,eAA4B,IAAI,IAAI,OAAO,cAAc;EAE/D,IAAI,CAAC,MAAM,mBAAmB,YAAY,GACxC,MAAM,IAAI,MACR,6BAA6B,CAAC,GAAG,iBAAiB,EAAE,YAClD,OAAO,gBAEX;EAEF,OAAO,IAAI,KAAK;GAAE;GAAK;GAAQ,GAAG,QAAQ;EAAW,CAAC;CACxD;CAEA,MAAM,SACJ,UACA,aACA,OAAqB;EACnB,aAAa;EACb,WAAW;EACX,eAAe;CACjB,GACsB;EACtB,MAAM,SAAS,SAAS,KAAK,SAAsB,OAAe;GAChE,OAAO,QAAQ,KAAK;GACpB,QAAQ,QAAQ,KAAK;GACrB,QAAQ,YAAY,EAAE,CAAC,KAAK;EAC9B,EAAE;EAEF,OAAO,MAAM,KAAK,MAAM,MAAM;CAChC;AACF"}

package/dist/evaluation/qa/index.cjs

@@ -1 +1,2 @@
-require("./eval_chain.cjs");
+const require_eval_chain = require("./eval_chain.cjs");
+exports.QAEvalChain = require_eval_chain.QAEvalChain;

package/dist/evaluation/qa/index.d.cts

@@ -1 +1,2 @@
-import { QAEvalChain } from "./eval_chain.cjs";
+import { QAEvalChain } from "./eval_chain.cjs";
+export { QAEvalChain };

package/dist/evaluation/qa/index.d.ts

@@ -1 +1,2 @@
-import { QAEvalChain } from "./eval_chain.js";
+import { QAEvalChain } from "./eval_chain.js";
+export { QAEvalChain };

package/dist/evaluation/qa/index.js

@@ -1,2 +1,2 @@
-import "./eval_chain.js";
-export {};
+import { QAEvalChain } from "./eval_chain.js";
+export { QAEvalChain };

package/dist/evaluation/qa/prompt.cjs

@@ -1,5 +1,4 @@
-require("../../_virtual/_rolldown/runtime.cjs");
-const QA_PROMPT = /* @__PURE__ */ new (require("@langchain/core/prompts")).PromptTemplate({
+const QA_PROMPT = /* #__PURE__ */ new (require("@langchain/core/prompts")).PromptTemplate({
 	inputVariables: [
 		"query",
 		"result",

package/dist/evaluation/qa/prompt.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"prompt.cjs","names":["PromptTemplate"],"sources":["../../../src/evaluation/qa/prompt.ts"],"sourcesContent":["import { PromptTemplate } from \"@langchain/core/prompts\";\n\nconst QA_TEMPLATE = `You are a teacher grading a quiz.\nYou are given a question, the student's answer, and the true answer, and are asked to score the student answer as either CORRECT or INCORRECT.\n\nExample Format:\nQUESTION: question here\nSTUDENT ANSWER: student's answer here\nTRUE ANSWER: true answer here\nGRADE: CORRECT or INCORRECT here\n\nGrade the student answers based ONLY on their factual accuracy. Ignore differences in punctuation and phrasing between the student answer and true answer. It is OK if the student answer contains more information than the true answer, as long as it does not contain any conflicting statements. Begin! \n\nQUESTION: {query}\nSTUDENT ANSWER: {result}\nTRUE ANSWER: {answer}\nGRADE:`;\nexport const QA_PROMPT = /* #__PURE__ */ new PromptTemplate({\n  inputVariables: [\"query\", \"result\", \"answer\"],\n  template: QA_TEMPLATE,\n});\n\nconst SQL_TEMPLATE = `You are comparing a submitted answer to an expert answer on a given SQL coding question. Here is the data:\n[BEGIN DATA]\n***\n[Question]: {query}\n***\n[Expert]: {answer}\n***\n[Submission]: {result}\n***\n[END DATA]\nCompare the content and correctness of the submitted SQL with the expert answer. Ignore any differences in whitespace, style, or output column names. The submitted answer may either be correct or incorrect. Determine which case applies. First, explain in detail the similarities or differences between the expert answer and the submission, ignoring superficial aspects such as whitespace, style or output column names. Do not state the final answer in your initial explanation. Then, respond with either \"CORRECT\" or \"INCORRECT\" (without quotes or punctuation) on its own line. This should correspond to whether the submitted SQL and the expert answer are semantically the same or different, respectively. Then, repeat your final answer on a new line.`;\n\nexport const SQL_PROMPT = /* #__PURE__ */ new PromptTemplate({\n  inputVariables: [\"query\", \"answer\", \"result\"],\n  template: SQL_TEMPLATE,\n});\n"],"mappings":";AAiBA,MAAa,4BAA4B,uCAAIA,EAAAA,eAAe;CAC1D,gBAAgB;EAAC;EAAS;EAAU;EAAS;CAC7C,UAAU;;;;;;;;;;;;;;;CACX,CAAC"}
+{"version":3,"file":"prompt.cjs","names":["PromptTemplate"],"sources":["../../../src/evaluation/qa/prompt.ts"],"sourcesContent":["import { PromptTemplate } from \"@langchain/core/prompts\";\n\nconst QA_TEMPLATE = `You are a teacher grading a quiz.\nYou are given a question, the student's answer, and the true answer, and are asked to score the student answer as either CORRECT or INCORRECT.\n\nExample Format:\nQUESTION: question here\nSTUDENT ANSWER: student's answer here\nTRUE ANSWER: true answer here\nGRADE: CORRECT or INCORRECT here\n\nGrade the student answers based ONLY on their factual accuracy. Ignore differences in punctuation and phrasing between the student answer and true answer. It is OK if the student answer contains more information than the true answer, as long as it does not contain any conflicting statements. Begin! \n\nQUESTION: {query}\nSTUDENT ANSWER: {result}\nTRUE ANSWER: {answer}\nGRADE:`;\nexport const QA_PROMPT = /* #__PURE__ */ new PromptTemplate({\n  inputVariables: [\"query\", \"result\", \"answer\"],\n  template: QA_TEMPLATE,\n});\n\nconst SQL_TEMPLATE = `You are comparing a submitted answer to an expert answer on a given SQL coding question. Here is the data:\n[BEGIN DATA]\n***\n[Question]: {query}\n***\n[Expert]: {answer}\n***\n[Submission]: {result}\n***\n[END DATA]\nCompare the content and correctness of the submitted SQL with the expert answer. Ignore any differences in whitespace, style, or output column names. The submitted answer may either be correct or incorrect. Determine which case applies. First, explain in detail the similarities or differences between the expert answer and the submission, ignoring superficial aspects such as whitespace, style or output column names. Do not state the final answer in your initial explanation. Then, respond with either \"CORRECT\" or \"INCORRECT\" (without quotes or punctuation) on its own line. This should correspond to whether the submitted SQL and the expert answer are semantically the same or different, respectively. Then, repeat your final answer on a new line.`;\n\nexport const SQL_PROMPT = /* #__PURE__ */ new PromptTemplate({\n  inputVariables: [\"query\", \"answer\", \"result\"],\n  template: SQL_TEMPLATE,\n});\n"],"mappings":"AAiBA,MAAa,4BAA4B,sCAAIA,EAAAA,CAAAA,eAAe;CAC1D,gBAAgB;EAAC;EAAS;EAAU;CAAQ;CAC5C,UAAU;;;;;;;;;;;;;;;AACZ,CAAC"}

package/dist/evaluation/qa/prompt.js

@@ -1,5 +1,5 @@
 import { PromptTemplate } from "@langchain/core/prompts";
-const QA_PROMPT = /* @__PURE__ */ new PromptTemplate({
+const QA_PROMPT = /* #__PURE__ */ new PromptTemplate({
 	inputVariables: [
 		"query",
 		"result",

package/dist/evaluation/qa/prompt.js.map

@@ -1 +1 @@
-{"version":3,"file":"prompt.js","names":[],"sources":["../../../src/evaluation/qa/prompt.ts"],"sourcesContent":["import { PromptTemplate } from \"@langchain/core/prompts\";\n\nconst QA_TEMPLATE = `You are a teacher grading a quiz.\nYou are given a question, the student's answer, and the true answer, and are asked to score the student answer as either CORRECT or INCORRECT.\n\nExample Format:\nQUESTION: question here\nSTUDENT ANSWER: student's answer here\nTRUE ANSWER: true answer here\nGRADE: CORRECT or INCORRECT here\n\nGrade the student answers based ONLY on their factual accuracy. Ignore differences in punctuation and phrasing between the student answer and true answer. It is OK if the student answer contains more information than the true answer, as long as it does not contain any conflicting statements. Begin! \n\nQUESTION: {query}\nSTUDENT ANSWER: {result}\nTRUE ANSWER: {answer}\nGRADE:`;\nexport const QA_PROMPT = /* #__PURE__ */ new PromptTemplate({\n  inputVariables: [\"query\", \"result\", \"answer\"],\n  template: QA_TEMPLATE,\n});\n\nconst SQL_TEMPLATE = `You are comparing a submitted answer to an expert answer on a given SQL coding question. Here is the data:\n[BEGIN DATA]\n***\n[Question]: {query}\n***\n[Expert]: {answer}\n***\n[Submission]: {result}\n***\n[END DATA]\nCompare the content and correctness of the submitted SQL with the expert answer. Ignore any differences in whitespace, style, or output column names. The submitted answer may either be correct or incorrect. Determine which case applies. First, explain in detail the similarities or differences between the expert answer and the submission, ignoring superficial aspects such as whitespace, style or output column names. Do not state the final answer in your initial explanation. Then, respond with either \"CORRECT\" or \"INCORRECT\" (without quotes or punctuation) on its own line. This should correspond to whether the submitted SQL and the expert answer are semantically the same or different, respectively. Then, repeat your final answer on a new line.`;\n\nexport const SQL_PROMPT = /* #__PURE__ */ new PromptTemplate({\n  inputVariables: [\"query\", \"answer\", \"result\"],\n  template: SQL_TEMPLATE,\n});\n"],"mappings":";AAiBA,MAAa,4BAA4B,IAAI,eAAe;CAC1D,gBAAgB;EAAC;EAAS;EAAU;EAAS;CAC7C,UAAU;;;;;;;;;;;;;;;CACX,CAAC"}
+{"version":3,"file":"prompt.js","names":[],"sources":["../../../src/evaluation/qa/prompt.ts"],"sourcesContent":["import { PromptTemplate } from \"@langchain/core/prompts\";\n\nconst QA_TEMPLATE = `You are a teacher grading a quiz.\nYou are given a question, the student's answer, and the true answer, and are asked to score the student answer as either CORRECT or INCORRECT.\n\nExample Format:\nQUESTION: question here\nSTUDENT ANSWER: student's answer here\nTRUE ANSWER: true answer here\nGRADE: CORRECT or INCORRECT here\n\nGrade the student answers based ONLY on their factual accuracy. Ignore differences in punctuation and phrasing between the student answer and true answer. It is OK if the student answer contains more information than the true answer, as long as it does not contain any conflicting statements. Begin! \n\nQUESTION: {query}\nSTUDENT ANSWER: {result}\nTRUE ANSWER: {answer}\nGRADE:`;\nexport const QA_PROMPT = /* #__PURE__ */ new PromptTemplate({\n  inputVariables: [\"query\", \"result\", \"answer\"],\n  template: QA_TEMPLATE,\n});\n\nconst SQL_TEMPLATE = `You are comparing a submitted answer to an expert answer on a given SQL coding question. Here is the data:\n[BEGIN DATA]\n***\n[Question]: {query}\n***\n[Expert]: {answer}\n***\n[Submission]: {result}\n***\n[END DATA]\nCompare the content and correctness of the submitted SQL with the expert answer. Ignore any differences in whitespace, style, or output column names. The submitted answer may either be correct or incorrect. Determine which case applies. First, explain in detail the similarities or differences between the expert answer and the submission, ignoring superficial aspects such as whitespace, style or output column names. Do not state the final answer in your initial explanation. Then, respond with either \"CORRECT\" or \"INCORRECT\" (without quotes or punctuation) on its own line. This should correspond to whether the submitted SQL and the expert answer are semantically the same or different, respectively. Then, repeat your final answer on a new line.`;\n\nexport const SQL_PROMPT = /* #__PURE__ */ new PromptTemplate({\n  inputVariables: [\"query\", \"answer\", \"result\"],\n  template: SQL_TEMPLATE,\n});\n"],"mappings":";AAiBA,MAAa,4BAA4B,IAAI,eAAe;CAC1D,gBAAgB;EAAC;EAAS;EAAU;CAAQ;CAC5C,UAAU;;;;;;;;;;;;;;;AACZ,CAAC"}

package/dist/evaluation/types.d.cts

@@ -1,5 +1,4 @@
 import { AgentTrajectoryEvaluator, LLMPairwiseStringEvaluator, LLMStringEvaluator, PairwiseStringEvaluator, StringEvaluator } from "./base.cjs";
-
 //#region src/evaluation/types.d.ts
 interface EvaluatorType {
   /**

package/dist/evaluation/types.d.cts.map

@@ -1 +1 @@
-{"version":3,"file":"types.d.cts","names":[],"sources":["../../src/evaluation/types.ts"],"mappings":";;;UAQiB,aAAA;;AAAjB;;;EAKE,QAAA,EAAU,kBAAA;EAOQ;;;;EAAlB,gBAAA,EAAkB,kBAAA;EA2BW;;;;EArB7B,eAAA,EAAiB,0BAAA;EANjB;;;;EAYA,uBAAA,EAAyB,0BAAA;EAAA;;;EAKzB,UAAA,EAAY,wBAAA;EAKQ;;;EAApB,kBAAA,EAAoB,eAAA;EAKgC;;;EAApD,2BAAA,EAA6B,uBAAA;AAAA"}
+{"version":3,"file":"types.d.cts","names":[],"sources":["../../src/evaluation/types.ts"],"mappings":";;UAQiB;;;;;EAKf,UAAU;;;;;EAOV,kBAAkB;;;;;EAMlB,iBAAiB;;;;;EAMjB,yBAAyB;;;;EAKzB,YAAY;;;;EAKZ,oBAAoB;;;;EAKpB,6BAA6B"}

package/dist/evaluation/types.d.ts

@@ -1,5 +1,4 @@
 import { AgentTrajectoryEvaluator, LLMPairwiseStringEvaluator, LLMStringEvaluator, PairwiseStringEvaluator, StringEvaluator } from "./base.js";
-
 //#region src/evaluation/types.d.ts
 interface EvaluatorType {
   /**

package/dist/evaluation/types.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"types.d.ts","names":[],"sources":["../../src/evaluation/types.ts"],"mappings":";;;UAQiB,aAAA;;AAAjB;;;EAKE,QAAA,EAAU,kBAAA;EAOQ;;;;EAAlB,gBAAA,EAAkB,kBAAA;EA2BW;;;;EArB7B,eAAA,EAAiB,0BAAA;EANjB;;;;EAYA,uBAAA,EAAyB,0BAAA;EAAA;;;EAKzB,UAAA,EAAY,wBAAA;EAKQ;;;EAApB,kBAAA,EAAoB,eAAA;EAKgC;;;EAApD,2BAAA,EAA6B,uBAAA;AAAA"}
+{"version":3,"file":"types.d.ts","names":[],"sources":["../../src/evaluation/types.ts"],"mappings":";;UAQiB;;;;;EAKf,UAAU;;;;;EAOV,kBAAkB;;;;;EAMlB,iBAAiB;;;;;EAMjB,yBAAyB;;;;EAKzB,YAAY;;;;EAKZ,oBAAoB;;;;EAKpB,6BAA6B"}

package/dist/experimental/autogpt/agent.cjs

@@ -1,4 +1,3 @@
-require("../../_virtual/_rolldown/runtime.cjs");
 const require_llm_chain = require("../../chains/llm_chain.cjs");
 require("./schema.cjs");
 const require_prompt = require("./prompt.cjs");
@@ -72,19 +71,20 @@ var AutoGPT = class AutoGPT {
 	* @returns A new instance of the AutoGPT class.
 	*/
 	static fromLLMAndTools(llm, tools, { aiName, aiRole, memory, maxIterations = 100, outputParser = new require_output_parser.AutoGPTOutputParser() }) {
+		const chain = new require_llm_chain.LLMChain({
+			llm,
+			prompt: new require_prompt.AutoGPTPrompt({
+				aiName,
+				aiRole,
+				tools,
+				tokenCounter: llm.getNumTokens.bind(llm),
+				sendTokenLimit: (0, _langchain_core_language_models_base.getModelContextSize)("modelName" in llm ? llm.modelName : "gpt2")
+			})
+		});
 		return new AutoGPT({
 			aiName,
 			memory,
-			chain: new require_llm_chain.LLMChain({
-				llm,
-				prompt: new require_prompt.AutoGPTPrompt({
-					aiName,
-					aiRole,
-					tools,
-					tokenCounter: llm.getNumTokens.bind(llm),
-					sendTokenLimit: (0, _langchain_core_language_models_base.getModelContextSize)("modelName" in llm ? llm.modelName : "gpt2")
-				})
-			}),
+			chain,
 			outputParser,
 			tools,
 			maxIterations

package/dist/experimental/autogpt/agent.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"agent.cjs","names":["AutoGPTOutputParser","LLMChain","AutoGPTPrompt","HumanMessage","AIMessage","SystemMessage"],"sources":["../../../src/experimental/autogpt/agent.ts"],"sourcesContent":["import type { VectorStoreRetrieverInterface } from \"@langchain/core/vectorstores\";\nimport { Tool } from \"@langchain/core/tools\";\nimport {\n  AIMessage,\n  BaseMessage,\n  HumanMessage,\n  SystemMessage,\n} from \"@langchain/core/messages\";\nimport { BaseChatModel } from \"@langchain/core/language_models/chat_models\";\nimport {\n  getEmbeddingContextSize,\n  getModelContextSize,\n} from \"@langchain/core/language_models/base\";\nimport { LLMChain } from \"../../chains/llm_chain.js\";\n\nimport { AutoGPTOutputParser } from \"./output_parser.js\";\nimport { AutoGPTPrompt } from \"./prompt.js\";\n// import { HumanInputRun } from \"./tools/human/tool\"; // TODO\nimport { ObjectTool, FINISH_NAME } from \"./schema.js\";\nimport { TokenTextSplitter } from \"../../text_splitter.js\";\n\n/**\n * Interface for the input parameters of the AutoGPT class.\n */\nexport interface AutoGPTInput {\n  aiName: string;\n  aiRole: string;\n  memory: VectorStoreRetrieverInterface;\n  humanInTheLoop?: boolean;\n  outputParser?: AutoGPTOutputParser;\n  maxIterations?: number;\n}\n\n/**\n * Class representing the AutoGPT concept with LangChain primitives. It is\n * designed to be used with a set of tools such as a search tool,\n * write-file tool, and a read-file tool.\n * @example\n * ```typescript\n * const autogpt = AutoGPT.fromLLMAndTools(\n *   new ChatOpenAI({ model: \"gpt-4o-mini\", temperature: 0 }),\n *   [\n *     new ReadFileTool({ store: new InMemoryFileStore() }),\n *     new WriteFileTool({ store: new InMemoryFileStore() }),\n *     new SerpAPI(\"YOUR_SERPAPI_API_KEY\", {\n *       location: \"San Francisco,California,United States\",\n *       hl: \"en\",\n *       gl: \"us\",\n *     }),\n *   ],\n *   {\n *     memory: new MemoryVectorStore(new OpenAIEmbeddings()).asRetriever(),\n *     aiName: \"Tom\",\n *     aiRole: \"Assistant\",\n *   },\n * );\n * const result = await autogpt.run([\"write a weather report for SF today\"]);\n * ```\n */\nexport class AutoGPT {\n  aiName: string;\n\n  memory: VectorStoreRetrieverInterface;\n\n  fullMessageHistory: BaseMessage[];\n\n  nextActionCount: number;\n\n  chain: LLMChain;\n\n  outputParser: AutoGPTOutputParser;\n\n  tools: ObjectTool[];\n\n  feedbackTool?: Tool;\n\n  maxIterations: number;\n\n  // Currently not generic enough to support any text splitter.\n  textSplitter: TokenTextSplitter;\n\n  constructor({\n    aiName,\n    memory,\n    chain,\n    outputParser,\n    tools,\n    feedbackTool,\n    maxIterations,\n  }: Omit<Required<AutoGPTInput>, \"aiRole\" | \"humanInTheLoop\"> & {\n    chain: LLMChain;\n    tools: ObjectTool[];\n    feedbackTool?: Tool;\n  }) {\n    this.aiName = aiName;\n    this.memory = memory;\n    this.fullMessageHistory = [];\n    this.nextActionCount = 0;\n    this.chain = chain;\n    this.outputParser = outputParser;\n    this.tools = tools;\n    this.feedbackTool = feedbackTool;\n    this.maxIterations = maxIterations;\n    const chunkSize = getEmbeddingContextSize(\n      \"modelName\" in memory.vectorStore.embeddings\n        ? (memory.vectorStore.embeddings.modelName as string)\n        : undefined\n    );\n    this.textSplitter = new TokenTextSplitter({\n      chunkSize,\n      chunkOverlap: Math.round(chunkSize / 10),\n    });\n  }\n\n  /**\n   * Creates a new AutoGPT instance from a given LLM and a set of tools.\n   * @param llm A BaseChatModel object.\n   * @param tools An array of ObjectTool objects.\n   * @param options.aiName The name of the AI.\n   * @param options.aiRole The role of the AI.\n   * @param options.memory A VectorStoreRetriever object that represents the memory of the AI.\n   * @param options.maxIterations The maximum number of iterations the AI can perform.\n   * @param options.outputParser An AutoGPTOutputParser object that parses the output of the AI.\n   * @returns A new instance of the AutoGPT class.\n   */\n  static fromLLMAndTools(\n    llm: BaseChatModel,\n    tools: ObjectTool[],\n    {\n      aiName,\n      aiRole,\n      memory,\n      maxIterations = 100,\n      // humanInTheLoop = false,\n      outputParser = new AutoGPTOutputParser(),\n    }: AutoGPTInput\n  ): AutoGPT {\n    const prompt = new AutoGPTPrompt({\n      aiName,\n      aiRole,\n      tools,\n      tokenCounter: llm.getNumTokens.bind(llm),\n      sendTokenLimit: getModelContextSize(\n        \"modelName\" in llm ? (llm.modelName as string) : \"gpt2\"\n      ),\n    });\n    // const feedbackTool = humanInTheLoop ? new HumanInputRun() : null;\n    const chain = new LLMChain({ llm, prompt });\n    return new AutoGPT({\n      aiName,\n      memory,\n      chain,\n      outputParser,\n      tools,\n      // feedbackTool,\n      maxIterations,\n    });\n  }\n\n  /**\n   * Runs the AI with a given set of goals.\n   * @param goals An array of strings representing the goals.\n   * @returns A string representing the result of the run or undefined if the maximum number of iterations is reached without a result.\n   */\n  async run(goals: string[]): Promise<string | undefined> {\n    const user_input =\n      \"Determine which next command to use, and respond using the format specified above:\";\n    let loopCount = 0;\n    while (loopCount < this.maxIterations) {\n      loopCount += 1;\n\n      const { text: assistantReply } = await this.chain.call({\n        goals,\n        user_input,\n        memory: this.memory,\n        messages: this.fullMessageHistory,\n      });\n\n      // Print the assistant reply\n      console.log(assistantReply);\n      this.fullMessageHistory.push(new HumanMessage(user_input));\n      this.fullMessageHistory.push(new AIMessage(assistantReply));\n\n      const action = await this.outputParser.parse(assistantReply);\n      const tools = this.tools.reduce(\n        (acc, tool) => ({ ...acc, [tool.name]: tool }),\n        {} as { [key: string]: ObjectTool }\n      );\n      if (action.name === FINISH_NAME) {\n        return action.args.response;\n      }\n      let result: string;\n      if (action.name in tools) {\n        const tool = tools[action.name];\n        let observation;\n        try {\n          observation = await tool.call(action.args);\n        } catch (e) {\n          observation = `Error in args: ${e}`;\n        }\n        result = `Command ${tool.name} returned: ${observation}`;\n      } else if (action.name === \"ERROR\") {\n        result = `Error: ${action.args}. `;\n      } else {\n        result = `Unknown command '${action.name}'. Please refer to the 'COMMANDS' list for available commands and only respond in the specified JSON format.`;\n      }\n\n      let memoryToAdd = `Assistant Reply: ${assistantReply}\\nResult: ${result} `;\n      if (this.feedbackTool) {\n        const feedback = `\\n${await this.feedbackTool.call(\"Input: \")}`;\n        if (feedback === \"q\" || feedback === \"stop\") {\n          console.log(\"EXITING\");\n          return \"EXITING\";\n        }\n        memoryToAdd += feedback;\n      }\n\n      const documents = await this.textSplitter.createDocuments([memoryToAdd]);\n      await this.memory.addDocuments(documents);\n      this.fullMessageHistory.push(new SystemMessage(result));\n    }\n\n    return undefined;\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA2DA,IAAa,UAAb,MAAa,QAAQ;CACnB;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAGA;CAEA,YAAY,EACV,QACA,QACA,OACA,cACA,OACA,cACA,iBAKC;AACD,OAAK,SAAS;AACd,OAAK,SAAS;AACd,OAAK,qBAAqB,EAAE;AAC5B,OAAK,kBAAkB;AACvB,OAAK,QAAQ;AACb,OAAK,eAAe;AACpB,OAAK,QAAQ;AACb,OAAK,eAAe;AACpB,OAAK,gBAAgB;EACrB,MAAM,aAAA,GAAA,qCAAA,yBACJ,eAAe,OAAO,YAAY,aAC7B,OAAO,YAAY,WAAW,YAC/B,KAAA,EACL;AACD,OAAK,eAAe,IAAA,sBAAA,sBAAI,kBAAkB;GACxC;GACA,cAAc,KAAK,MAAM,YAAY,GAAG;GACzC,CAAC;;;;;;;;;;;;;CAcJ,OAAO,gBACL,KACA,OACA,EACE,QACA,QACA,QACA,gBAAgB,KAEhB,eAAe,IAAIA,sBAAAA,qBAAqB,IAEjC;AAYT,SAAO,IAAI,QAAQ;GACjB;GACA;GACA,OAAA,IAJgBC,kBAAAA,SAAS;IAAE;IAAK,QAAA,IAVfC,eAAAA,cAAc;KAC/B;KACA;KACA;KACA,cAAc,IAAI,aAAa,KAAK,IAAI;KACxC,iBAAA,GAAA,qCAAA,qBACE,eAAe,MAAO,IAAI,YAAuB,OAClD;KACF,CAEuC;IAAE,CAInC;GACL;GACA;GAEA;GACD,CAAC;;;;;;;CAQJ,MAAM,IAAI,OAA8C;EACtD,MAAM,aACJ;EACF,IAAI,YAAY;AAChB,SAAO,YAAY,KAAK,eAAe;AACrC,gBAAa;GAEb,MAAM,EAAE,MAAM,mBAAmB,MAAM,KAAK,MAAM,KAAK;IACrD;IACA;IACA,QAAQ,KAAK;IACb,UAAU,KAAK;IAChB,CAAC;AAGF,WAAQ,IAAI,eAAe;AAC3B,QAAK,mBAAmB,KAAK,IAAIC,yBAAAA,aAAa,WAAW,CAAC;AAC1D,QAAK,mBAAmB,KAAK,IAAIC,yBAAAA,UAAU,eAAe,CAAC;GAE3D,MAAM,SAAS,MAAM,KAAK,aAAa,MAAM,eAAe;GAC5D,MAAM,QAAQ,KAAK,MAAM,QACtB,KAAK,UAAU;IAAE,GAAG;KAAM,KAAK,OAAO;IAAM,GAC7C,EAAE,CACH;AACD,OAAI,OAAO,SAAA,SACT,QAAO,OAAO,KAAK;GAErB,IAAI;AACJ,OAAI,OAAO,QAAQ,OAAO;IACxB,MAAM,OAAO,MAAM,OAAO;IAC1B,IAAI;AACJ,QAAI;AACF,mBAAc,MAAM,KAAK,KAAK,OAAO,KAAK;aACnC,GAAG;AACV,mBAAc,kBAAkB;;AAElC,aAAS,WAAW,KAAK,KAAK,aAAa;cAClC,OAAO,SAAS,QACzB,UAAS,UAAU,OAAO,KAAK;OAE/B,UAAS,oBAAoB,OAAO,KAAK;GAG3C,IAAI,cAAc,oBAAoB,eAAe,YAAY,OAAO;AACxE,OAAI,KAAK,cAAc;IACrB,MAAM,WAAW,KAAK,MAAM,KAAK,aAAa,KAAK,UAAU;AAC7D,QAAI,aAAa,OAAO,aAAa,QAAQ;AAC3C,aAAQ,IAAI,UAAU;AACtB,YAAO;;AAET,mBAAe;;GAGjB,MAAM,YAAY,MAAM,KAAK,aAAa,gBAAgB,CAAC,YAAY,CAAC;AACxE,SAAM,KAAK,OAAO,aAAa,UAAU;AACzC,QAAK,mBAAmB,KAAK,IAAIC,yBAAAA,cAAc,OAAO,CAAC"}
+{"version":3,"file":"agent.cjs","names":["AutoGPTOutputParser","LLMChain","AutoGPTPrompt","HumanMessage","AIMessage","SystemMessage"],"sources":["../../../src/experimental/autogpt/agent.ts"],"sourcesContent":["import type { VectorStoreRetrieverInterface } from \"@langchain/core/vectorstores\";\nimport { Tool } from \"@langchain/core/tools\";\nimport {\n  AIMessage,\n  BaseMessage,\n  HumanMessage,\n  SystemMessage,\n} from \"@langchain/core/messages\";\nimport { BaseChatModel } from \"@langchain/core/language_models/chat_models\";\nimport {\n  getEmbeddingContextSize,\n  getModelContextSize,\n} from \"@langchain/core/language_models/base\";\nimport { LLMChain } from \"../../chains/llm_chain.js\";\n\nimport { AutoGPTOutputParser } from \"./output_parser.js\";\nimport { AutoGPTPrompt } from \"./prompt.js\";\n// import { HumanInputRun } from \"./tools/human/tool\"; // TODO\nimport { ObjectTool, FINISH_NAME } from \"./schema.js\";\nimport { TokenTextSplitter } from \"../../text_splitter.js\";\n\n/**\n * Interface for the input parameters of the AutoGPT class.\n */\nexport interface AutoGPTInput {\n  aiName: string;\n  aiRole: string;\n  memory: VectorStoreRetrieverInterface;\n  humanInTheLoop?: boolean;\n  outputParser?: AutoGPTOutputParser;\n  maxIterations?: number;\n}\n\n/**\n * Class representing the AutoGPT concept with LangChain primitives. It is\n * designed to be used with a set of tools such as a search tool,\n * write-file tool, and a read-file tool.\n * @example\n * ```typescript\n * const autogpt = AutoGPT.fromLLMAndTools(\n *   new ChatOpenAI({ model: \"gpt-4o-mini\", temperature: 0 }),\n *   [\n *     new ReadFileTool({ store: new InMemoryFileStore() }),\n *     new WriteFileTool({ store: new InMemoryFileStore() }),\n *     new SerpAPI(\"YOUR_SERPAPI_API_KEY\", {\n *       location: \"San Francisco,California,United States\",\n *       hl: \"en\",\n *       gl: \"us\",\n *     }),\n *   ],\n *   {\n *     memory: new MemoryVectorStore(new OpenAIEmbeddings()).asRetriever(),\n *     aiName: \"Tom\",\n *     aiRole: \"Assistant\",\n *   },\n * );\n * const result = await autogpt.run([\"write a weather report for SF today\"]);\n * ```\n */\nexport class AutoGPT {\n  aiName: string;\n\n  memory: VectorStoreRetrieverInterface;\n\n  fullMessageHistory: BaseMessage[];\n\n  nextActionCount: number;\n\n  chain: LLMChain;\n\n  outputParser: AutoGPTOutputParser;\n\n  tools: ObjectTool[];\n\n  feedbackTool?: Tool;\n\n  maxIterations: number;\n\n  // Currently not generic enough to support any text splitter.\n  textSplitter: TokenTextSplitter;\n\n  constructor({\n    aiName,\n    memory,\n    chain,\n    outputParser,\n    tools,\n    feedbackTool,\n    maxIterations,\n  }: Omit<Required<AutoGPTInput>, \"aiRole\" | \"humanInTheLoop\"> & {\n    chain: LLMChain;\n    tools: ObjectTool[];\n    feedbackTool?: Tool;\n  }) {\n    this.aiName = aiName;\n    this.memory = memory;\n    this.fullMessageHistory = [];\n    this.nextActionCount = 0;\n    this.chain = chain;\n    this.outputParser = outputParser;\n    this.tools = tools;\n    this.feedbackTool = feedbackTool;\n    this.maxIterations = maxIterations;\n    const chunkSize = getEmbeddingContextSize(\n      \"modelName\" in memory.vectorStore.embeddings\n        ? (memory.vectorStore.embeddings.modelName as string)\n        : undefined\n    );\n    this.textSplitter = new TokenTextSplitter({\n      chunkSize,\n      chunkOverlap: Math.round(chunkSize / 10),\n    });\n  }\n\n  /**\n   * Creates a new AutoGPT instance from a given LLM and a set of tools.\n   * @param llm A BaseChatModel object.\n   * @param tools An array of ObjectTool objects.\n   * @param options.aiName The name of the AI.\n   * @param options.aiRole The role of the AI.\n   * @param options.memory A VectorStoreRetriever object that represents the memory of the AI.\n   * @param options.maxIterations The maximum number of iterations the AI can perform.\n   * @param options.outputParser An AutoGPTOutputParser object that parses the output of the AI.\n   * @returns A new instance of the AutoGPT class.\n   */\n  static fromLLMAndTools(\n    llm: BaseChatModel,\n    tools: ObjectTool[],\n    {\n      aiName,\n      aiRole,\n      memory,\n      maxIterations = 100,\n      // humanInTheLoop = false,\n      outputParser = new AutoGPTOutputParser(),\n    }: AutoGPTInput\n  ): AutoGPT {\n    const prompt = new AutoGPTPrompt({\n      aiName,\n      aiRole,\n      tools,\n      tokenCounter: llm.getNumTokens.bind(llm),\n      sendTokenLimit: getModelContextSize(\n        \"modelName\" in llm ? (llm.modelName as string) : \"gpt2\"\n      ),\n    });\n    // const feedbackTool = humanInTheLoop ? new HumanInputRun() : null;\n    const chain = new LLMChain({ llm, prompt });\n    return new AutoGPT({\n      aiName,\n      memory,\n      chain,\n      outputParser,\n      tools,\n      // feedbackTool,\n      maxIterations,\n    });\n  }\n\n  /**\n   * Runs the AI with a given set of goals.\n   * @param goals An array of strings representing the goals.\n   * @returns A string representing the result of the run or undefined if the maximum number of iterations is reached without a result.\n   */\n  async run(goals: string[]): Promise<string | undefined> {\n    const user_input =\n      \"Determine which next command to use, and respond using the format specified above:\";\n    let loopCount = 0;\n    while (loopCount < this.maxIterations) {\n      loopCount += 1;\n\n      const { text: assistantReply } = await this.chain.call({\n        goals,\n        user_input,\n        memory: this.memory,\n        messages: this.fullMessageHistory,\n      });\n\n      // Print the assistant reply\n      console.log(assistantReply);\n      this.fullMessageHistory.push(new HumanMessage(user_input));\n      this.fullMessageHistory.push(new AIMessage(assistantReply));\n\n      const action = await this.outputParser.parse(assistantReply);\n      const tools = this.tools.reduce(\n        (acc, tool) => ({ ...acc, [tool.name]: tool }),\n        {} as { [key: string]: ObjectTool }\n      );\n      if (action.name === FINISH_NAME) {\n        return action.args.response;\n      }\n      let result: string;\n      if (action.name in tools) {\n        const tool = tools[action.name];\n        let observation;\n        try {\n          observation = await tool.call(action.args);\n        } catch (e) {\n          observation = `Error in args: ${e}`;\n        }\n        result = `Command ${tool.name} returned: ${observation}`;\n      } else if (action.name === \"ERROR\") {\n        result = `Error: ${action.args}. `;\n      } else {\n        result = `Unknown command '${action.name}'. Please refer to the 'COMMANDS' list for available commands and only respond in the specified JSON format.`;\n      }\n\n      let memoryToAdd = `Assistant Reply: ${assistantReply}\\nResult: ${result} `;\n      if (this.feedbackTool) {\n        const feedback = `\\n${await this.feedbackTool.call(\"Input: \")}`;\n        if (feedback === \"q\" || feedback === \"stop\") {\n          console.log(\"EXITING\");\n          return \"EXITING\";\n        }\n        memoryToAdd += feedback;\n      }\n\n      const documents = await this.textSplitter.createDocuments([memoryToAdd]);\n      await this.memory.addDocuments(documents);\n      this.fullMessageHistory.push(new SystemMessage(result));\n    }\n\n    return undefined;\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA2DA,IAAa,UAAb,MAAa,QAAQ;CACnB;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAGA;CAEA,YAAY,EACV,QACA,QACA,OACA,cACA,OACA,cACA,iBAKC;EACD,KAAK,SAAS;EACd,KAAK,SAAS;EACd,KAAK,qBAAqB,CAAC;EAC3B,KAAK,kBAAkB;EACvB,KAAK,QAAQ;EACb,KAAK,eAAe;EACpB,KAAK,QAAQ;EACb,KAAK,eAAe;EACpB,KAAK,gBAAgB;EACrB,MAAM,aAAA,GAAA,qCAAA,wBAAA,CACJ,eAAe,OAAO,YAAY,aAC7B,OAAO,YAAY,WAAW,YAC/B,KAAA,CACN;EACA,KAAK,eAAe,IAAA,sBAAA,sBAAI,kBAAkB;GACxC;GACA,cAAc,KAAK,MAAM,YAAY,EAAE;EACzC,CAAC;CACH;;;;;;;;;;;;CAaA,OAAO,gBACL,KACA,OACA,EACE,QACA,QACA,QACA,gBAAgB,KAEhB,eAAe,IAAIA,sBAAAA,oBAAoB,KAEhC;EAWT,MAAM,QAAQ,IAAIC,kBAAAA,SAAS;GAAE;GAAK,QAAA,IAVfC,eAAAA,cAAc;IAC/B;IACA;IACA;IACA,cAAc,IAAI,aAAa,KAAK,GAAG;IACvC,iBAAA,GAAA,qCAAA,oBAAA,CACE,eAAe,MAAO,IAAI,YAAuB,MACnD;GACF,CAEuC;EAAE,CAAC;EAC1C,OAAO,IAAI,QAAQ;GACjB;GACA;GACA;GACA;GACA;GAEA;EACF,CAAC;CACH;;;;;;CAOA,MAAM,IAAI,OAA8C;EACtD,MAAM,aACJ;EACF,IAAI,YAAY;EAChB,OAAO,YAAY,KAAK,eAAe;GACrC,aAAa;GAEb,MAAM,EAAE,MAAM,mBAAmB,MAAM,KAAK,MAAM,KAAK;IACrD;IACA;IACA,QAAQ,KAAK;IACb,UAAU,KAAK;GACjB,CAAC;GAGD,QAAQ,IAAI,cAAc;GAC1B,KAAK,mBAAmB,KAAK,IAAIC,yBAAAA,aAAa,UAAU,CAAC;GACzD,KAAK,mBAAmB,KAAK,IAAIC,yBAAAA,UAAU,cAAc,CAAC;GAE1D,MAAM,SAAS,MAAM,KAAK,aAAa,MAAM,cAAc;GAC3D,MAAM,QAAQ,KAAK,MAAM,QACtB,KAAK,UAAU;IAAE,GAAG;KAAM,KAAK,OAAO;GAAK,IAC5C,CAAC,CACH;GACA,IAAI,OAAO,SAAA,UACT,OAAO,OAAO,KAAK;GAErB,IAAI;GACJ,IAAI,OAAO,QAAQ,OAAO;IACxB,MAAM,OAAO,MAAM,OAAO;IAC1B,IAAI;IACJ,IAAI;KACF,cAAc,MAAM,KAAK,KAAK,OAAO,IAAI;IAC3C,SAAS,GAAG;KACV,cAAc,kBAAkB;IAClC;IACA,SAAS,WAAW,KAAK,KAAK,aAAa;GAC7C,OAAO,IAAI,OAAO,SAAS,SACzB,SAAS,UAAU,OAAO,KAAK;QAE/B,SAAS,oBAAoB,OAAO,KAAK;GAG3C,IAAI,cAAc,oBAAoB,eAAe,YAAY,OAAO;GACxE,IAAI,KAAK,cAAc;IACrB,MAAM,WAAW,KAAK,MAAM,KAAK,aAAa,KAAK,SAAS;IAC5D,IAAI,aAAa,OAAO,aAAa,QAAQ;KAC3C,QAAQ,IAAI,SAAS;KACrB,OAAO;IACT;IACA,eAAe;GACjB;GAEA,MAAM,YAAY,MAAM,KAAK,aAAa,gBAAgB,CAAC,WAAW,CAAC;GACvE,MAAM,KAAK,OAAO,aAAa,SAAS;GACxC,KAAK,mBAAmB,KAAK,IAAIC,yBAAAA,cAAc,MAAM,CAAC;EACxD;CAGF;AACF"}

package/dist/experimental/autogpt/agent.d.cts

@@ -6,7 +6,6 @@ import { BaseMessage } from "@langchain/core/messages";
 import { Tool } from "@langchain/core/tools";
 import { VectorStoreRetrieverInterface } from "@langchain/core/vectorstores";
 import { BaseChatModel } from "@langchain/core/language_models/chat_models";
-
 //#region src/experimental/autogpt/agent.d.ts
 /**
  * Interface for the input parameters of the AutoGPT class.
@@ -56,15 +55,7 @@ declare class AutoGPT {
   feedbackTool?: Tool;
   maxIterations: number;
   textSplitter: text_splitter_d_exports.TokenTextSplitter;
-  constructor({
-    aiName,
-    memory,
-    chain,
-    outputParser,
-    tools,
-    feedbackTool,
-    maxIterations
-  }: Omit<Required<AutoGPTInput>, "aiRole" | "humanInTheLoop"> & {
+  constructor({ aiName, memory, chain, outputParser, tools, feedbackTool, maxIterations }: Omit<Required<AutoGPTInput>, "aiRole" | "humanInTheLoop"> & {
     chain: LLMChain;
     tools: ObjectTool[];
     feedbackTool?: Tool;
@@ -80,13 +71,7 @@ declare class AutoGPT {
    * @param options.outputParser An AutoGPTOutputParser object that parses the output of the AI.
    * @returns A new instance of the AutoGPT class.
    */
-  static fromLLMAndTools(llm: BaseChatModel, tools: ObjectTool[], {
-    aiName,
-    aiRole,
-    memory,
-    maxIterations,
-    outputParser
-  }: AutoGPTInput): AutoGPT;
+  static fromLLMAndTools(llm: BaseChatModel, tools: ObjectTool[], { aiName, aiRole, memory, maxIterations, outputParser }: AutoGPTInput): AutoGPT;
   /**
    * Runs the AI with a given set of goals.
    * @param goals An array of strings representing the goals.

package/dist/experimental/autogpt/agent.d.cts.map

@@ -1 +1 @@
-{"version":3,"file":"agent.d.cts","names":[],"sources":["../../../src/experimental/autogpt/agent.ts"],"mappings":";;;;;;;;;;;;AAwBA;UAAiB,YAAA;EACf,MAAA;EACA,MAAA;EACA,MAAA,EAAQ,6BAAA;EACR,cAAA;EACA,YAAA,GAAe,mBAAA;EACf,aAAA;AAAA;;;;;;AA6BF;;;;;;;;;;;;;;;;;;;;;cAAa,OAAA;EACX,MAAA;EAEA,MAAA,EAAQ,6BAAA;EAER,kBAAA,EAAoB,WAAA;EAEpB,eAAA;EAEA,KAAA,EAAO,QAAA;EAEP,YAAA,EAAc,mBAAA;EAEd,KAAA,EAAO,UAAA;EAEP,YAAA,GAAe,IAAA;EAEf,aAAA;EAGA,YAAA,EAAc,uBAAA,CAAA,iBAAA;EAEd,WAAA,CAAA;IACE,MAAA;IACA,MAAA;IACA,KAAA;IACA,YAAA;IACA,KAAA;IACA,YAAA;IACA;EAAA,GACC,IAAA,CAAK,QAAA,CAAS,YAAA;IACf,KAAA,EAAO,QAAA;IACP,KAAA,EAAO,UAAA;IACP,YAAA,GAAe,IAAA;EAAA;EA9BT;;;;;;;;;;;EAAA,OA+DD,eAAA,CACL,GAAA,EAAK,aAAA,EACL,KAAA,EAAO,UAAA;IAEL,MAAA;IACA,MAAA;IACA,MAAA;IACA,aAAA;IAEA;EAAA,GACC,YAAA,GACF,OAAA;EA5DH;;;;;EAwFM,GAAA,CAAI,KAAA,aAAkB,OAAA;AAAA"}
+{"version":3,"file":"agent.d.cts","names":[],"sources":["../../../src/experimental/autogpt/agent.ts"],"mappings":";;;;;;;;;;;;UAwBiB;EACf;EACA;EACA,QAAQ;EACR;EACA,eAAe;EACf;;;;;;;;;;;;;;;;;;;;;;;;;;;;cA6BW;EACX;EAEA,QAAQ;EAER,oBAAoB;EAEpB;EAEA,OAAO;EAEP,cAAc;EAEd,OAAO;EAEP,eAAe;EAEf;EAGA,cAAc,wBAAA;EAEd,cACE,QACA,QACA,OACA,cACA,OACA,cACA,iBACC,KAAK,SAAS;IACf,OAAO;IACP,OAAO;IACP,eAAe;;;;;;;;;;;;;SAiCV,gBACL,KAAK,eACL,OAAO,gBAEL,QACA,QACA,QACA,eAEA,gBACC,eACF;;;;;;EA4BG,IAAI,kBAAkB"}

package/dist/experimental/autogpt/agent.d.ts

@@ -6,7 +6,6 @@ import { BaseMessage } from "@langchain/core/messages";
 import { Tool } from "@langchain/core/tools";
 import { BaseChatModel } from "@langchain/core/language_models/chat_models";
 import { VectorStoreRetrieverInterface } from "@langchain/core/vectorstores";
-
 //#region src/experimental/autogpt/agent.d.ts
 /**
  * Interface for the input parameters of the AutoGPT class.
@@ -56,15 +55,7 @@ declare class AutoGPT {
   feedbackTool?: Tool;
   maxIterations: number;
   textSplitter: text_splitter_d_exports.TokenTextSplitter;
-  constructor({
-    aiName,
-    memory,
-    chain,
-    outputParser,
-    tools,
-    feedbackTool,
-    maxIterations
-  }: Omit<Required<AutoGPTInput>, "aiRole" | "humanInTheLoop"> & {
+  constructor({ aiName, memory, chain, outputParser, tools, feedbackTool, maxIterations }: Omit<Required<AutoGPTInput>, "aiRole" | "humanInTheLoop"> & {
     chain: LLMChain;
     tools: ObjectTool[];
     feedbackTool?: Tool;
@@ -80,13 +71,7 @@ declare class AutoGPT {
    * @param options.outputParser An AutoGPTOutputParser object that parses the output of the AI.
    * @returns A new instance of the AutoGPT class.
    */
-  static fromLLMAndTools(llm: BaseChatModel, tools: ObjectTool[], {
-    aiName,
-    aiRole,
-    memory,
-    maxIterations,
-    outputParser
-  }: AutoGPTInput): AutoGPT;
+  static fromLLMAndTools(llm: BaseChatModel, tools: ObjectTool[], { aiName, aiRole, memory, maxIterations, outputParser }: AutoGPTInput): AutoGPT;
   /**
    * Runs the AI with a given set of goals.
    * @param goals An array of strings representing the goals.

package/dist/experimental/autogpt/agent.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"agent.d.ts","names":[],"sources":["../../../src/experimental/autogpt/agent.ts"],"mappings":";;;;;;;;;;;;AAwBA;UAAiB,YAAA;EACf,MAAA;EACA,MAAA;EACA,MAAA,EAAQ,6BAAA;EACR,cAAA;EACA,YAAA,GAAe,mBAAA;EACf,aAAA;AAAA;;;;;;AA6BF;;;;;;;;;;;;;;;;;;;;;cAAa,OAAA;EACX,MAAA;EAEA,MAAA,EAAQ,6BAAA;EAER,kBAAA,EAAoB,WAAA;EAEpB,eAAA;EAEA,KAAA,EAAO,QAAA;EAEP,YAAA,EAAc,mBAAA;EAEd,KAAA,EAAO,UAAA;EAEP,YAAA,GAAe,IAAA;EAEf,aAAA;EAGA,YAAA,EAAc,uBAAA,CAAA,iBAAA;EAEd,WAAA,CAAA;IACE,MAAA;IACA,MAAA;IACA,KAAA;IACA,YAAA;IACA,KAAA;IACA,YAAA;IACA;EAAA,GACC,IAAA,CAAK,QAAA,CAAS,YAAA;IACf,KAAA,EAAO,QAAA;IACP,KAAA,EAAO,UAAA;IACP,YAAA,GAAe,IAAA;EAAA;EA9BT;;;;;;;;;;;EAAA,OA+DD,eAAA,CACL,GAAA,EAAK,aAAA,EACL,KAAA,EAAO,UAAA;IAEL,MAAA;IACA,MAAA;IACA,MAAA;IACA,aAAA;IAEA;EAAA,GACC,YAAA,GACF,OAAA;EA5DH;;;;;EAwFM,GAAA,CAAI,KAAA,aAAkB,OAAA;AAAA"}
+{"version":3,"file":"agent.d.ts","names":[],"sources":["../../../src/experimental/autogpt/agent.ts"],"mappings":";;;;;;;;;;;;UAwBiB;EACf;EACA;EACA,QAAQ;EACR;EACA,eAAe;EACf;;;;;;;;;;;;;;;;;;;;;;;;;;;;cA6BW;EACX;EAEA,QAAQ;EAER,oBAAoB;EAEpB;EAEA,OAAO;EAEP,cAAc;EAEd,OAAO;EAEP,eAAe;EAEf;EAGA,cAAc,wBAAA;EAEd,cACE,QACA,QACA,OACA,cACA,OACA,cACA,iBACC,KAAK,SAAS;IACf,OAAO;IACP,OAAO;IACP,eAAe;;;;;;;;;;;;;SAiCV,gBACL,KAAK,eACL,OAAO,gBAEL,QACA,QACA,QACA,eAEA,gBACC,eACF;;;;;;EA4BG,IAAI,kBAAkB"}

package/dist/experimental/autogpt/agent.js

@@ -71,19 +71,20 @@ var AutoGPT = class AutoGPT {
 	* @returns A new instance of the AutoGPT class.
 	*/
 	static fromLLMAndTools(llm, tools, { aiName, aiRole, memory, maxIterations = 100, outputParser = new AutoGPTOutputParser() }) {
+		const chain = new LLMChain({
+			llm,
+			prompt: new AutoGPTPrompt({
+				aiName,
+				aiRole,
+				tools,
+				tokenCounter: llm.getNumTokens.bind(llm),
+				sendTokenLimit: getModelContextSize("modelName" in llm ? llm.modelName : "gpt2")
+			})
+		});
 		return new AutoGPT({
 			aiName,
 			memory,
-			chain: new LLMChain({
-				llm,
-				prompt: new AutoGPTPrompt({
-					aiName,
-					aiRole,
-					tools,
-					tokenCounter: llm.getNumTokens.bind(llm),
-					sendTokenLimit: getModelContextSize("modelName" in llm ? llm.modelName : "gpt2")
-				})
-			}),
+			chain,
 			outputParser,
 			tools,
 			maxIterations

package/dist/experimental/autogpt/agent.js.map

@@ -1 +1 @@
-{"version":3,"file":"agent.js","names":["TokenTextSplitter"],"sources":["../../../src/experimental/autogpt/agent.ts"],"sourcesContent":["import type { VectorStoreRetrieverInterface } from \"@langchain/core/vectorstores\";\nimport { Tool } from \"@langchain/core/tools\";\nimport {\n  AIMessage,\n  BaseMessage,\n  HumanMessage,\n  SystemMessage,\n} from \"@langchain/core/messages\";\nimport { BaseChatModel } from \"@langchain/core/language_models/chat_models\";\nimport {\n  getEmbeddingContextSize,\n  getModelContextSize,\n} from \"@langchain/core/language_models/base\";\nimport { LLMChain } from \"../../chains/llm_chain.js\";\n\nimport { AutoGPTOutputParser } from \"./output_parser.js\";\nimport { AutoGPTPrompt } from \"./prompt.js\";\n// import { HumanInputRun } from \"./tools/human/tool\"; // TODO\nimport { ObjectTool, FINISH_NAME } from \"./schema.js\";\nimport { TokenTextSplitter } from \"../../text_splitter.js\";\n\n/**\n * Interface for the input parameters of the AutoGPT class.\n */\nexport interface AutoGPTInput {\n  aiName: string;\n  aiRole: string;\n  memory: VectorStoreRetrieverInterface;\n  humanInTheLoop?: boolean;\n  outputParser?: AutoGPTOutputParser;\n  maxIterations?: number;\n}\n\n/**\n * Class representing the AutoGPT concept with LangChain primitives. It is\n * designed to be used with a set of tools such as a search tool,\n * write-file tool, and a read-file tool.\n * @example\n * ```typescript\n * const autogpt = AutoGPT.fromLLMAndTools(\n *   new ChatOpenAI({ model: \"gpt-4o-mini\", temperature: 0 }),\n *   [\n *     new ReadFileTool({ store: new InMemoryFileStore() }),\n *     new WriteFileTool({ store: new InMemoryFileStore() }),\n *     new SerpAPI(\"YOUR_SERPAPI_API_KEY\", {\n *       location: \"San Francisco,California,United States\",\n *       hl: \"en\",\n *       gl: \"us\",\n *     }),\n *   ],\n *   {\n *     memory: new MemoryVectorStore(new OpenAIEmbeddings()).asRetriever(),\n *     aiName: \"Tom\",\n *     aiRole: \"Assistant\",\n *   },\n * );\n * const result = await autogpt.run([\"write a weather report for SF today\"]);\n * ```\n */\nexport class AutoGPT {\n  aiName: string;\n\n  memory: VectorStoreRetrieverInterface;\n\n  fullMessageHistory: BaseMessage[];\n\n  nextActionCount: number;\n\n  chain: LLMChain;\n\n  outputParser: AutoGPTOutputParser;\n\n  tools: ObjectTool[];\n\n  feedbackTool?: Tool;\n\n  maxIterations: number;\n\n  // Currently not generic enough to support any text splitter.\n  textSplitter: TokenTextSplitter;\n\n  constructor({\n    aiName,\n    memory,\n    chain,\n    outputParser,\n    tools,\n    feedbackTool,\n    maxIterations,\n  }: Omit<Required<AutoGPTInput>, \"aiRole\" | \"humanInTheLoop\"> & {\n    chain: LLMChain;\n    tools: ObjectTool[];\n    feedbackTool?: Tool;\n  }) {\n    this.aiName = aiName;\n    this.memory = memory;\n    this.fullMessageHistory = [];\n    this.nextActionCount = 0;\n    this.chain = chain;\n    this.outputParser = outputParser;\n    this.tools = tools;\n    this.feedbackTool = feedbackTool;\n    this.maxIterations = maxIterations;\n    const chunkSize = getEmbeddingContextSize(\n      \"modelName\" in memory.vectorStore.embeddings\n        ? (memory.vectorStore.embeddings.modelName as string)\n        : undefined\n    );\n    this.textSplitter = new TokenTextSplitter({\n      chunkSize,\n      chunkOverlap: Math.round(chunkSize / 10),\n    });\n  }\n\n  /**\n   * Creates a new AutoGPT instance from a given LLM and a set of tools.\n   * @param llm A BaseChatModel object.\n   * @param tools An array of ObjectTool objects.\n   * @param options.aiName The name of the AI.\n   * @param options.aiRole The role of the AI.\n   * @param options.memory A VectorStoreRetriever object that represents the memory of the AI.\n   * @param options.maxIterations The maximum number of iterations the AI can perform.\n   * @param options.outputParser An AutoGPTOutputParser object that parses the output of the AI.\n   * @returns A new instance of the AutoGPT class.\n   */\n  static fromLLMAndTools(\n    llm: BaseChatModel,\n    tools: ObjectTool[],\n    {\n      aiName,\n      aiRole,\n      memory,\n      maxIterations = 100,\n      // humanInTheLoop = false,\n      outputParser = new AutoGPTOutputParser(),\n    }: AutoGPTInput\n  ): AutoGPT {\n    const prompt = new AutoGPTPrompt({\n      aiName,\n      aiRole,\n      tools,\n      tokenCounter: llm.getNumTokens.bind(llm),\n      sendTokenLimit: getModelContextSize(\n        \"modelName\" in llm ? (llm.modelName as string) : \"gpt2\"\n      ),\n    });\n    // const feedbackTool = humanInTheLoop ? new HumanInputRun() : null;\n    const chain = new LLMChain({ llm, prompt });\n    return new AutoGPT({\n      aiName,\n      memory,\n      chain,\n      outputParser,\n      tools,\n      // feedbackTool,\n      maxIterations,\n    });\n  }\n\n  /**\n   * Runs the AI with a given set of goals.\n   * @param goals An array of strings representing the goals.\n   * @returns A string representing the result of the run or undefined if the maximum number of iterations is reached without a result.\n   */\n  async run(goals: string[]): Promise<string | undefined> {\n    const user_input =\n      \"Determine which next command to use, and respond using the format specified above:\";\n    let loopCount = 0;\n    while (loopCount < this.maxIterations) {\n      loopCount += 1;\n\n      const { text: assistantReply } = await this.chain.call({\n        goals,\n        user_input,\n        memory: this.memory,\n        messages: this.fullMessageHistory,\n      });\n\n      // Print the assistant reply\n      console.log(assistantReply);\n      this.fullMessageHistory.push(new HumanMessage(user_input));\n      this.fullMessageHistory.push(new AIMessage(assistantReply));\n\n      const action = await this.outputParser.parse(assistantReply);\n      const tools = this.tools.reduce(\n        (acc, tool) => ({ ...acc, [tool.name]: tool }),\n        {} as { [key: string]: ObjectTool }\n      );\n      if (action.name === FINISH_NAME) {\n        return action.args.response;\n      }\n      let result: string;\n      if (action.name in tools) {\n        const tool = tools[action.name];\n        let observation;\n        try {\n          observation = await tool.call(action.args);\n        } catch (e) {\n          observation = `Error in args: ${e}`;\n        }\n        result = `Command ${tool.name} returned: ${observation}`;\n      } else if (action.name === \"ERROR\") {\n        result = `Error: ${action.args}. `;\n      } else {\n        result = `Unknown command '${action.name}'. Please refer to the 'COMMANDS' list for available commands and only respond in the specified JSON format.`;\n      }\n\n      let memoryToAdd = `Assistant Reply: ${assistantReply}\\nResult: ${result} `;\n      if (this.feedbackTool) {\n        const feedback = `\\n${await this.feedbackTool.call(\"Input: \")}`;\n        if (feedback === \"q\" || feedback === \"stop\") {\n          console.log(\"EXITING\");\n          return \"EXITING\";\n        }\n        memoryToAdd += feedback;\n      }\n\n      const documents = await this.textSplitter.createDocuments([memoryToAdd]);\n      await this.memory.addDocuments(documents);\n      this.fullMessageHistory.push(new SystemMessage(result));\n    }\n\n    return undefined;\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA2DA,IAAa,UAAb,MAAa,QAAQ;CACnB;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAGA;CAEA,YAAY,EACV,QACA,QACA,OACA,cACA,OACA,cACA,iBAKC;AACD,OAAK,SAAS;AACd,OAAK,SAAS;AACd,OAAK,qBAAqB,EAAE;AAC5B,OAAK,kBAAkB;AACvB,OAAK,QAAQ;AACb,OAAK,eAAe;AACpB,OAAK,QAAQ;AACb,OAAK,eAAe;AACpB,OAAK,gBAAgB;EACrB,MAAM,YAAY,wBAChB,eAAe,OAAO,YAAY,aAC7B,OAAO,YAAY,WAAW,YAC/B,KAAA,EACL;AACD,OAAK,eAAe,IAAIA,sBAAAA,kBAAkB;GACxC;GACA,cAAc,KAAK,MAAM,YAAY,GAAG;GACzC,CAAC;;;;;;;;;;;;;CAcJ,OAAO,gBACL,KACA,OACA,EACE,QACA,QACA,QACA,gBAAgB,KAEhB,eAAe,IAAI,qBAAqB,IAEjC;AAYT,SAAO,IAAI,QAAQ;GACjB;GACA;GACA,OAAA,IAJgB,SAAS;IAAE;IAAK,QAAA,IAVf,cAAc;KAC/B;KACA;KACA;KACA,cAAc,IAAI,aAAa,KAAK,IAAI;KACxC,gBAAgB,oBACd,eAAe,MAAO,IAAI,YAAuB,OAClD;KACF,CAEuC;IAAE,CAInC;GACL;GACA;GAEA;GACD,CAAC;;;;;;;CAQJ,MAAM,IAAI,OAA8C;EACtD,MAAM,aACJ;EACF,IAAI,YAAY;AAChB,SAAO,YAAY,KAAK,eAAe;AACrC,gBAAa;GAEb,MAAM,EAAE,MAAM,mBAAmB,MAAM,KAAK,MAAM,KAAK;IACrD;IACA;IACA,QAAQ,KAAK;IACb,UAAU,KAAK;IAChB,CAAC;AAGF,WAAQ,IAAI,eAAe;AAC3B,QAAK,mBAAmB,KAAK,IAAI,aAAa,WAAW,CAAC;AAC1D,QAAK,mBAAmB,KAAK,IAAI,UAAU,eAAe,CAAC;GAE3D,MAAM,SAAS,MAAM,KAAK,aAAa,MAAM,eAAe;GAC5D,MAAM,QAAQ,KAAK,MAAM,QACtB,KAAK,UAAU;IAAE,GAAG;KAAM,KAAK,OAAO;IAAM,GAC7C,EAAE,CACH;AACD,OAAI,OAAO,SAAA,SACT,QAAO,OAAO,KAAK;GAErB,IAAI;AACJ,OAAI,OAAO,QAAQ,OAAO;IACxB,MAAM,OAAO,MAAM,OAAO;IAC1B,IAAI;AACJ,QAAI;AACF,mBAAc,MAAM,KAAK,KAAK,OAAO,KAAK;aACnC,GAAG;AACV,mBAAc,kBAAkB;;AAElC,aAAS,WAAW,KAAK,KAAK,aAAa;cAClC,OAAO,SAAS,QACzB,UAAS,UAAU,OAAO,KAAK;OAE/B,UAAS,oBAAoB,OAAO,KAAK;GAG3C,IAAI,cAAc,oBAAoB,eAAe,YAAY,OAAO;AACxE,OAAI,KAAK,cAAc;IACrB,MAAM,WAAW,KAAK,MAAM,KAAK,aAAa,KAAK,UAAU;AAC7D,QAAI,aAAa,OAAO,aAAa,QAAQ;AAC3C,aAAQ,IAAI,UAAU;AACtB,YAAO;;AAET,mBAAe;;GAGjB,MAAM,YAAY,MAAM,KAAK,aAAa,gBAAgB,CAAC,YAAY,CAAC;AACxE,SAAM,KAAK,OAAO,aAAa,UAAU;AACzC,QAAK,mBAAmB,KAAK,IAAI,cAAc,OAAO,CAAC"}
+{"version":3,"file":"agent.js","names":["TokenTextSplitter"],"sources":["../../../src/experimental/autogpt/agent.ts"],"sourcesContent":["import type { VectorStoreRetrieverInterface } from \"@langchain/core/vectorstores\";\nimport { Tool } from \"@langchain/core/tools\";\nimport {\n  AIMessage,\n  BaseMessage,\n  HumanMessage,\n  SystemMessage,\n} from \"@langchain/core/messages\";\nimport { BaseChatModel } from \"@langchain/core/language_models/chat_models\";\nimport {\n  getEmbeddingContextSize,\n  getModelContextSize,\n} from \"@langchain/core/language_models/base\";\nimport { LLMChain } from \"../../chains/llm_chain.js\";\n\nimport { AutoGPTOutputParser } from \"./output_parser.js\";\nimport { AutoGPTPrompt } from \"./prompt.js\";\n// import { HumanInputRun } from \"./tools/human/tool\"; // TODO\nimport { ObjectTool, FINISH_NAME } from \"./schema.js\";\nimport { TokenTextSplitter } from \"../../text_splitter.js\";\n\n/**\n * Interface for the input parameters of the AutoGPT class.\n */\nexport interface AutoGPTInput {\n  aiName: string;\n  aiRole: string;\n  memory: VectorStoreRetrieverInterface;\n  humanInTheLoop?: boolean;\n  outputParser?: AutoGPTOutputParser;\n  maxIterations?: number;\n}\n\n/**\n * Class representing the AutoGPT concept with LangChain primitives. It is\n * designed to be used with a set of tools such as a search tool,\n * write-file tool, and a read-file tool.\n * @example\n * ```typescript\n * const autogpt = AutoGPT.fromLLMAndTools(\n *   new ChatOpenAI({ model: \"gpt-4o-mini\", temperature: 0 }),\n *   [\n *     new ReadFileTool({ store: new InMemoryFileStore() }),\n *     new WriteFileTool({ store: new InMemoryFileStore() }),\n *     new SerpAPI(\"YOUR_SERPAPI_API_KEY\", {\n *       location: \"San Francisco,California,United States\",\n *       hl: \"en\",\n *       gl: \"us\",\n *     }),\n *   ],\n *   {\n *     memory: new MemoryVectorStore(new OpenAIEmbeddings()).asRetriever(),\n *     aiName: \"Tom\",\n *     aiRole: \"Assistant\",\n *   },\n * );\n * const result = await autogpt.run([\"write a weather report for SF today\"]);\n * ```\n */\nexport class AutoGPT {\n  aiName: string;\n\n  memory: VectorStoreRetrieverInterface;\n\n  fullMessageHistory: BaseMessage[];\n\n  nextActionCount: number;\n\n  chain: LLMChain;\n\n  outputParser: AutoGPTOutputParser;\n\n  tools: ObjectTool[];\n\n  feedbackTool?: Tool;\n\n  maxIterations: number;\n\n  // Currently not generic enough to support any text splitter.\n  textSplitter: TokenTextSplitter;\n\n  constructor({\n    aiName,\n    memory,\n    chain,\n    outputParser,\n    tools,\n    feedbackTool,\n    maxIterations,\n  }: Omit<Required<AutoGPTInput>, \"aiRole\" | \"humanInTheLoop\"> & {\n    chain: LLMChain;\n    tools: ObjectTool[];\n    feedbackTool?: Tool;\n  }) {\n    this.aiName = aiName;\n    this.memory = memory;\n    this.fullMessageHistory = [];\n    this.nextActionCount = 0;\n    this.chain = chain;\n    this.outputParser = outputParser;\n    this.tools = tools;\n    this.feedbackTool = feedbackTool;\n    this.maxIterations = maxIterations;\n    const chunkSize = getEmbeddingContextSize(\n      \"modelName\" in memory.vectorStore.embeddings\n        ? (memory.vectorStore.embeddings.modelName as string)\n        : undefined\n    );\n    this.textSplitter = new TokenTextSplitter({\n      chunkSize,\n      chunkOverlap: Math.round(chunkSize / 10),\n    });\n  }\n\n  /**\n   * Creates a new AutoGPT instance from a given LLM and a set of tools.\n   * @param llm A BaseChatModel object.\n   * @param tools An array of ObjectTool objects.\n   * @param options.aiName The name of the AI.\n   * @param options.aiRole The role of the AI.\n   * @param options.memory A VectorStoreRetriever object that represents the memory of the AI.\n   * @param options.maxIterations The maximum number of iterations the AI can perform.\n   * @param options.outputParser An AutoGPTOutputParser object that parses the output of the AI.\n   * @returns A new instance of the AutoGPT class.\n   */\n  static fromLLMAndTools(\n    llm: BaseChatModel,\n    tools: ObjectTool[],\n    {\n      aiName,\n      aiRole,\n      memory,\n      maxIterations = 100,\n      // humanInTheLoop = false,\n      outputParser = new AutoGPTOutputParser(),\n    }: AutoGPTInput\n  ): AutoGPT {\n    const prompt = new AutoGPTPrompt({\n      aiName,\n      aiRole,\n      tools,\n      tokenCounter: llm.getNumTokens.bind(llm),\n      sendTokenLimit: getModelContextSize(\n        \"modelName\" in llm ? (llm.modelName as string) : \"gpt2\"\n      ),\n    });\n    // const feedbackTool = humanInTheLoop ? new HumanInputRun() : null;\n    const chain = new LLMChain({ llm, prompt });\n    return new AutoGPT({\n      aiName,\n      memory,\n      chain,\n      outputParser,\n      tools,\n      // feedbackTool,\n      maxIterations,\n    });\n  }\n\n  /**\n   * Runs the AI with a given set of goals.\n   * @param goals An array of strings representing the goals.\n   * @returns A string representing the result of the run or undefined if the maximum number of iterations is reached without a result.\n   */\n  async run(goals: string[]): Promise<string | undefined> {\n    const user_input =\n      \"Determine which next command to use, and respond using the format specified above:\";\n    let loopCount = 0;\n    while (loopCount < this.maxIterations) {\n      loopCount += 1;\n\n      const { text: assistantReply } = await this.chain.call({\n        goals,\n        user_input,\n        memory: this.memory,\n        messages: this.fullMessageHistory,\n      });\n\n      // Print the assistant reply\n      console.log(assistantReply);\n      this.fullMessageHistory.push(new HumanMessage(user_input));\n      this.fullMessageHistory.push(new AIMessage(assistantReply));\n\n      const action = await this.outputParser.parse(assistantReply);\n      const tools = this.tools.reduce(\n        (acc, tool) => ({ ...acc, [tool.name]: tool }),\n        {} as { [key: string]: ObjectTool }\n      );\n      if (action.name === FINISH_NAME) {\n        return action.args.response;\n      }\n      let result: string;\n      if (action.name in tools) {\n        const tool = tools[action.name];\n        let observation;\n        try {\n          observation = await tool.call(action.args);\n        } catch (e) {\n          observation = `Error in args: ${e}`;\n        }\n        result = `Command ${tool.name} returned: ${observation}`;\n      } else if (action.name === \"ERROR\") {\n        result = `Error: ${action.args}. `;\n      } else {\n        result = `Unknown command '${action.name}'. Please refer to the 'COMMANDS' list for available commands and only respond in the specified JSON format.`;\n      }\n\n      let memoryToAdd = `Assistant Reply: ${assistantReply}\\nResult: ${result} `;\n      if (this.feedbackTool) {\n        const feedback = `\\n${await this.feedbackTool.call(\"Input: \")}`;\n        if (feedback === \"q\" || feedback === \"stop\") {\n          console.log(\"EXITING\");\n          return \"EXITING\";\n        }\n        memoryToAdd += feedback;\n      }\n\n      const documents = await this.textSplitter.createDocuments([memoryToAdd]);\n      await this.memory.addDocuments(documents);\n      this.fullMessageHistory.push(new SystemMessage(result));\n    }\n\n    return undefined;\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA2DA,IAAa,UAAb,MAAa,QAAQ;CACnB;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAEA;CAGA;CAEA,YAAY,EACV,QACA,QACA,OACA,cACA,OACA,cACA,iBAKC;EACD,KAAK,SAAS;EACd,KAAK,SAAS;EACd,KAAK,qBAAqB,CAAC;EAC3B,KAAK,kBAAkB;EACvB,KAAK,QAAQ;EACb,KAAK,eAAe;EACpB,KAAK,QAAQ;EACb,KAAK,eAAe;EACpB,KAAK,gBAAgB;EACrB,MAAM,YAAY,wBAChB,eAAe,OAAO,YAAY,aAC7B,OAAO,YAAY,WAAW,YAC/B,KAAA,CACN;EACA,KAAK,eAAe,IAAIA,sBAAAA,kBAAkB;GACxC;GACA,cAAc,KAAK,MAAM,YAAY,EAAE;EACzC,CAAC;CACH;;;;;;;;;;;;CAaA,OAAO,gBACL,KACA,OACA,EACE,QACA,QACA,QACA,gBAAgB,KAEhB,eAAe,IAAI,oBAAoB,KAEhC;EAWT,MAAM,QAAQ,IAAI,SAAS;GAAE;GAAK,QAAA,IAVf,cAAc;IAC/B;IACA;IACA;IACA,cAAc,IAAI,aAAa,KAAK,GAAG;IACvC,gBAAgB,oBACd,eAAe,MAAO,IAAI,YAAuB,MACnD;GACF,CAEuC;EAAE,CAAC;EAC1C,OAAO,IAAI,QAAQ;GACjB;GACA;GACA;GACA;GACA;GAEA;EACF,CAAC;CACH;;;;;;CAOA,MAAM,IAAI,OAA8C;EACtD,MAAM,aACJ;EACF,IAAI,YAAY;EAChB,OAAO,YAAY,KAAK,eAAe;GACrC,aAAa;GAEb,MAAM,EAAE,MAAM,mBAAmB,MAAM,KAAK,MAAM,KAAK;IACrD;IACA;IACA,QAAQ,KAAK;IACb,UAAU,KAAK;GACjB,CAAC;GAGD,QAAQ,IAAI,cAAc;GAC1B,KAAK,mBAAmB,KAAK,IAAI,aAAa,UAAU,CAAC;GACzD,KAAK,mBAAmB,KAAK,IAAI,UAAU,cAAc,CAAC;GAE1D,MAAM,SAAS,MAAM,KAAK,aAAa,MAAM,cAAc;GAC3D,MAAM,QAAQ,KAAK,MAAM,QACtB,KAAK,UAAU;IAAE,GAAG;KAAM,KAAK,OAAO;GAAK,IAC5C,CAAC,CACH;GACA,IAAI,OAAO,SAAA,UACT,OAAO,OAAO,KAAK;GAErB,IAAI;GACJ,IAAI,OAAO,QAAQ,OAAO;IACxB,MAAM,OAAO,MAAM,OAAO;IAC1B,IAAI;IACJ,IAAI;KACF,cAAc,MAAM,KAAK,KAAK,OAAO,IAAI;IAC3C,SAAS,GAAG;KACV,cAAc,kBAAkB;IAClC;IACA,SAAS,WAAW,KAAK,KAAK,aAAa;GAC7C,OAAO,IAAI,OAAO,SAAS,SACzB,SAAS,UAAU,OAAO,KAAK;QAE/B,SAAS,oBAAoB,OAAO,KAAK;GAG3C,IAAI,cAAc,oBAAoB,eAAe,YAAY,OAAO;GACxE,IAAI,KAAK,cAAc;IACrB,MAAM,WAAW,KAAK,MAAM,KAAK,aAAa,KAAK,SAAS;IAC5D,IAAI,aAAa,OAAO,aAAa,QAAQ;KAC3C,QAAQ,IAAI,SAAS;KACrB,OAAO;IACT;IACA,eAAe;GACjB;GAEA,MAAM,YAAY,MAAM,KAAK,aAAa,gBAAgB,CAAC,WAAW,CAAC;GACvE,MAAM,KAAK,OAAO,aAAa,SAAS;GACxC,KAAK,mBAAmB,KAAK,IAAI,cAAc,MAAM,CAAC;EACxD;CAGF;AACF"}

package/dist/experimental/autogpt/output_parser.cjs

@@ -1,4 +1,3 @@
-require("../../_virtual/_rolldown/runtime.cjs");
 let _langchain_core_output_parsers = require("@langchain/core/output_parsers");
 //#region src/experimental/autogpt/output_parser.ts
 /**

package/dist/experimental/autogpt/output_parser.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"output_parser.cjs","names":["BaseOutputParser"],"sources":["../../../src/experimental/autogpt/output_parser.ts"],"sourcesContent":["import { BaseOutputParser } from \"@langchain/core/output_parsers\";\nimport { AutoGPTAction } from \"./schema.js\";\n\n/**\n * Utility function used to preprocess a string to be parsed as JSON.\n * It replaces single backslashes with double backslashes, while leaving\n * already escaped ones intact.\n * It also extracts the json code if it is inside a code block\n */\nexport function preprocessJsonInput(inputStr: string): string {\n  const correctedStr = inputStr.replace(\n    /(?<!\\\\)\\\\(?![\"\\\\/bfnrt]|u[0-9a-fA-F]{4})/g,\n    \"\\\\\\\\\"\n  );\n  const match = correctedStr.match(\n    /```(.*)(\\r\\n|\\r|\\n)(?<code>[\\w\\W\\n]+)(\\r\\n|\\r|\\n)```/\n  );\n  if (match?.groups?.code) {\n    return match.groups.code.trim();\n  } else {\n    return correctedStr;\n  }\n}\n\n/**\n * Class responsible for parsing the output of AutoGPT. It extends the\n * BaseOutputParser class.\n */\nexport class AutoGPTOutputParser extends BaseOutputParser<AutoGPTAction> {\n  lc_namespace = [\"langchain\", \"experimental\", \"autogpt\"];\n\n  /**\n   * Method not implemented in the class and will throw an error if called.\n   * It is likely meant to be overridden in subclasses to provide specific\n   * format instructions.\n   * @returns Throws an error.\n   */\n  getFormatInstructions(): string {\n    throw new Error(\"Method not implemented.\");\n  }\n\n  /**\n   * Asynchronous method that takes a string as input and attempts to parse\n   * it into an AutoGPTAction object. If the input string cannot be parsed\n   * directly, the method tries to preprocess the string using the\n   * preprocessJsonInput function and parse it again. If parsing fails\n   * again, it returns an AutoGPTAction object with an error message.\n   * @param text The string to be parsed.\n   * @returns A Promise that resolves to an AutoGPTAction object.\n   */\n  async parse(text: string): Promise<AutoGPTAction> {\n    let parsed: {\n      command: {\n        name: string;\n        args: Record<string, unknown>;\n      };\n    };\n    try {\n      parsed = JSON.parse(text);\n    } catch {\n      const preprocessedText = preprocessJsonInput(text);\n      try {\n        parsed = JSON.parse(preprocessedText);\n      } catch {\n        return {\n          name: \"ERROR\",\n          args: { error: `Could not parse invalid json: ${text}` },\n        };\n      }\n    }\n    try {\n      return {\n        name: parsed.command.name,\n        args: parsed.command.args,\n      };\n    } catch {\n      return {\n        name: \"ERROR\",\n        args: { error: `Incomplete command args: ${parsed}` },\n      };\n    }\n  }\n}\n"],"mappings":";;;;;;;;;AASA,SAAgB,oBAAoB,UAA0B;CAC5D,MAAM,eAAe,SAAS,QAC5B,6CACA,OACD;CACD,MAAM,QAAQ,aAAa,MACzB,uDACD;AACD,KAAI,OAAO,QAAQ,KACjB,QAAO,MAAM,OAAO,KAAK,MAAM;KAE/B,QAAO;;;;;;AAQX,IAAa,sBAAb,cAAyCA,+BAAAA,iBAAgC;CACvE,eAAe;EAAC;EAAa;EAAgB;EAAU;;;;;;;CAQvD,wBAAgC;AAC9B,QAAM,IAAI,MAAM,0BAA0B;;;;;;;;;;;CAY5C,MAAM,MAAM,MAAsC;EAChD,IAAI;AAMJ,MAAI;AACF,YAAS,KAAK,MAAM,KAAK;UACnB;GACN,MAAM,mBAAmB,oBAAoB,KAAK;AAClD,OAAI;AACF,aAAS,KAAK,MAAM,iBAAiB;WAC/B;AACN,WAAO;KACL,MAAM;KACN,MAAM,EAAE,OAAO,iCAAiC,QAAQ;KACzD;;;AAGL,MAAI;AACF,UAAO;IACL,MAAM,OAAO,QAAQ;IACrB,MAAM,OAAO,QAAQ;IACtB;UACK;AACN,UAAO;IACL,MAAM;IACN,MAAM,EAAE,OAAO,4BAA4B,UAAU;IACtD"}
+{"version":3,"file":"output_parser.cjs","names":["BaseOutputParser"],"sources":["../../../src/experimental/autogpt/output_parser.ts"],"sourcesContent":["import { BaseOutputParser } from \"@langchain/core/output_parsers\";\nimport { AutoGPTAction } from \"./schema.js\";\n\n/**\n * Utility function used to preprocess a string to be parsed as JSON.\n * It replaces single backslashes with double backslashes, while leaving\n * already escaped ones intact.\n * It also extracts the json code if it is inside a code block\n */\nexport function preprocessJsonInput(inputStr: string): string {\n  const correctedStr = inputStr.replace(\n    /(?<!\\\\)\\\\(?![\"\\\\/bfnrt]|u[0-9a-fA-F]{4})/g,\n    \"\\\\\\\\\"\n  );\n  const match = correctedStr.match(\n    /```(.*)(\\r\\n|\\r|\\n)(?<code>[\\w\\W\\n]+)(\\r\\n|\\r|\\n)```/\n  );\n  if (match?.groups?.code) {\n    return match.groups.code.trim();\n  } else {\n    return correctedStr;\n  }\n}\n\n/**\n * Class responsible for parsing the output of AutoGPT. It extends the\n * BaseOutputParser class.\n */\nexport class AutoGPTOutputParser extends BaseOutputParser<AutoGPTAction> {\n  lc_namespace = [\"langchain\", \"experimental\", \"autogpt\"];\n\n  /**\n   * Method not implemented in the class and will throw an error if called.\n   * It is likely meant to be overridden in subclasses to provide specific\n   * format instructions.\n   * @returns Throws an error.\n   */\n  getFormatInstructions(): string {\n    throw new Error(\"Method not implemented.\");\n  }\n\n  /**\n   * Asynchronous method that takes a string as input and attempts to parse\n   * it into an AutoGPTAction object. If the input string cannot be parsed\n   * directly, the method tries to preprocess the string using the\n   * preprocessJsonInput function and parse it again. If parsing fails\n   * again, it returns an AutoGPTAction object with an error message.\n   * @param text The string to be parsed.\n   * @returns A Promise that resolves to an AutoGPTAction object.\n   */\n  async parse(text: string): Promise<AutoGPTAction> {\n    let parsed: {\n      command: {\n        name: string;\n        args: Record<string, unknown>;\n      };\n    };\n    try {\n      parsed = JSON.parse(text);\n    } catch {\n      const preprocessedText = preprocessJsonInput(text);\n      try {\n        parsed = JSON.parse(preprocessedText);\n      } catch {\n        return {\n          name: \"ERROR\",\n          args: { error: `Could not parse invalid json: ${text}` },\n        };\n      }\n    }\n    try {\n      return {\n        name: parsed.command.name,\n        args: parsed.command.args,\n      };\n    } catch {\n      return {\n        name: \"ERROR\",\n        args: { error: `Incomplete command args: ${parsed}` },\n      };\n    }\n  }\n}\n"],"mappings":";;;;;;;;AASA,SAAgB,oBAAoB,UAA0B;CAC5D,MAAM,eAAe,SAAS,QAC5B,6CACA,MACF;CACA,MAAM,QAAQ,aAAa,MACzB,sDACF;CACA,IAAI,OAAO,QAAQ,MACjB,OAAO,MAAM,OAAO,KAAK,KAAK;MAE9B,OAAO;AAEX;;;;;AAMA,IAAa,sBAAb,cAAyCA,+BAAAA,iBAAgC;CACvE,eAAe;EAAC;EAAa;EAAgB;CAAS;;;;;;;CAQtD,wBAAgC;EAC9B,MAAM,IAAI,MAAM,yBAAyB;CAC3C;;;;;;;;;;CAWA,MAAM,MAAM,MAAsC;EAChD,IAAI;EAMJ,IAAI;GACF,SAAS,KAAK,MAAM,IAAI;EAC1B,QAAQ;GACN,MAAM,mBAAmB,oBAAoB,IAAI;GACjD,IAAI;IACF,SAAS,KAAK,MAAM,gBAAgB;GACtC,QAAQ;IACN,OAAO;KACL,MAAM;KACN,MAAM,EAAE,OAAO,iCAAiC,OAAO;IACzD;GACF;EACF;EACA,IAAI;GACF,OAAO;IACL,MAAM,OAAO,QAAQ;IACrB,MAAM,OAAO,QAAQ;GACvB;EACF,QAAQ;GACN,OAAO;IACL,MAAM;IACN,MAAM,EAAE,OAAO,4BAA4B,SAAS;GACtD;EACF;CACF;AACF"}

package/dist/experimental/autogpt/output_parser.d.cts

@@ -1,6 +1,5 @@
 import { AutoGPTAction } from "./schema.cjs";
 import { BaseOutputParser } from "@langchain/core/output_parsers";
-
 //#region src/experimental/autogpt/output_parser.d.ts
 /**
  * Utility function used to preprocess a string to be parsed as JSON.

package/dist/experimental/autogpt/output_parser.d.cts.map

@@ -1 +1 @@
-{"version":3,"file":"output_parser.d.cts","names":[],"sources":["../../../src/experimental/autogpt/output_parser.ts"],"mappings":";;;;;;AASA;;;;iBAAgB,mBAAA,CAAoB,QAAA;AAmBpC;;;;AAAA,cAAa,mBAAA,SAA4B,gBAAA,CAAiB,aAAA;EACxD,YAAA;EADuC;;;;;;EASvC,qBAAA,CAAA;EAaM;;;;;;;;;EAAA,KAAA,CAAM,IAAA,WAAe,OAAA,CAAQ,aAAA;AAAA"}
+{"version":3,"file":"output_parser.d.cts","names":[],"sources":["../../../src/experimental/autogpt/output_parser.ts"],"mappings":";;;;;;;;;iBASgB,oBAAoB;;;;;cAmBvB,4BAA4B,iBAAiB;EACxD;;;;;;;EAQA;;;;;;;;;;EAaM,MAAM,eAAe,QAAQ"}

package/dist/experimental/autogpt/output_parser.d.ts

@@ -1,6 +1,5 @@
 import { AutoGPTAction } from "./schema.js";
 import { BaseOutputParser } from "@langchain/core/output_parsers";
-
 //#region src/experimental/autogpt/output_parser.d.ts
 /**
  * Utility function used to preprocess a string to be parsed as JSON.

package/dist/experimental/autogpt/output_parser.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"output_parser.d.ts","names":[],"sources":["../../../src/experimental/autogpt/output_parser.ts"],"mappings":";;;;;;AASA;;;;iBAAgB,mBAAA,CAAoB,QAAA;AAmBpC;;;;AAAA,cAAa,mBAAA,SAA4B,gBAAA,CAAiB,aAAA;EACxD,YAAA;EADuC;;;;;;EASvC,qBAAA,CAAA;EAaM;;;;;;;;;EAAA,KAAA,CAAM,IAAA,WAAe,OAAA,CAAQ,aAAA;AAAA"}
+{"version":3,"file":"output_parser.d.ts","names":[],"sources":["../../../src/experimental/autogpt/output_parser.ts"],"mappings":";;;;;;;;;iBASgB,oBAAoB;;;;;cAmBvB,4BAA4B,iBAAiB;EACxD;;;;;;;EAQA;;;;;;;;;;EAaM,MAAM,eAAe,QAAQ"}

package/dist/experimental/autogpt/output_parser.js.map

@@ -1 +1 @@
-{"version":3,"file":"output_parser.js","names":[],"sources":["../../../src/experimental/autogpt/output_parser.ts"],"sourcesContent":["import { BaseOutputParser } from \"@langchain/core/output_parsers\";\nimport { AutoGPTAction } from \"./schema.js\";\n\n/**\n * Utility function used to preprocess a string to be parsed as JSON.\n * It replaces single backslashes with double backslashes, while leaving\n * already escaped ones intact.\n * It also extracts the json code if it is inside a code block\n */\nexport function preprocessJsonInput(inputStr: string): string {\n  const correctedStr = inputStr.replace(\n    /(?<!\\\\)\\\\(?![\"\\\\/bfnrt]|u[0-9a-fA-F]{4})/g,\n    \"\\\\\\\\\"\n  );\n  const match = correctedStr.match(\n    /```(.*)(\\r\\n|\\r|\\n)(?<code>[\\w\\W\\n]+)(\\r\\n|\\r|\\n)```/\n  );\n  if (match?.groups?.code) {\n    return match.groups.code.trim();\n  } else {\n    return correctedStr;\n  }\n}\n\n/**\n * Class responsible for parsing the output of AutoGPT. It extends the\n * BaseOutputParser class.\n */\nexport class AutoGPTOutputParser extends BaseOutputParser<AutoGPTAction> {\n  lc_namespace = [\"langchain\", \"experimental\", \"autogpt\"];\n\n  /**\n   * Method not implemented in the class and will throw an error if called.\n   * It is likely meant to be overridden in subclasses to provide specific\n   * format instructions.\n   * @returns Throws an error.\n   */\n  getFormatInstructions(): string {\n    throw new Error(\"Method not implemented.\");\n  }\n\n  /**\n   * Asynchronous method that takes a string as input and attempts to parse\n   * it into an AutoGPTAction object. If the input string cannot be parsed\n   * directly, the method tries to preprocess the string using the\n   * preprocessJsonInput function and parse it again. If parsing fails\n   * again, it returns an AutoGPTAction object with an error message.\n   * @param text The string to be parsed.\n   * @returns A Promise that resolves to an AutoGPTAction object.\n   */\n  async parse(text: string): Promise<AutoGPTAction> {\n    let parsed: {\n      command: {\n        name: string;\n        args: Record<string, unknown>;\n      };\n    };\n    try {\n      parsed = JSON.parse(text);\n    } catch {\n      const preprocessedText = preprocessJsonInput(text);\n      try {\n        parsed = JSON.parse(preprocessedText);\n      } catch {\n        return {\n          name: \"ERROR\",\n          args: { error: `Could not parse invalid json: ${text}` },\n        };\n      }\n    }\n    try {\n      return {\n        name: parsed.command.name,\n        args: parsed.command.args,\n      };\n    } catch {\n      return {\n        name: \"ERROR\",\n        args: { error: `Incomplete command args: ${parsed}` },\n      };\n    }\n  }\n}\n"],"mappings":";;;;;;;;AASA,SAAgB,oBAAoB,UAA0B;CAC5D,MAAM,eAAe,SAAS,QAC5B,6CACA,OACD;CACD,MAAM,QAAQ,aAAa,MACzB,uDACD;AACD,KAAI,OAAO,QAAQ,KACjB,QAAO,MAAM,OAAO,KAAK,MAAM;KAE/B,QAAO;;;;;;AAQX,IAAa,sBAAb,cAAyC,iBAAgC;CACvE,eAAe;EAAC;EAAa;EAAgB;EAAU;;;;;;;CAQvD,wBAAgC;AAC9B,QAAM,IAAI,MAAM,0BAA0B;;;;;;;;;;;CAY5C,MAAM,MAAM,MAAsC;EAChD,IAAI;AAMJ,MAAI;AACF,YAAS,KAAK,MAAM,KAAK;UACnB;GACN,MAAM,mBAAmB,oBAAoB,KAAK;AAClD,OAAI;AACF,aAAS,KAAK,MAAM,iBAAiB;WAC/B;AACN,WAAO;KACL,MAAM;KACN,MAAM,EAAE,OAAO,iCAAiC,QAAQ;KACzD;;;AAGL,MAAI;AACF,UAAO;IACL,MAAM,OAAO,QAAQ;IACrB,MAAM,OAAO,QAAQ;IACtB;UACK;AACN,UAAO;IACL,MAAM;IACN,MAAM,EAAE,OAAO,4BAA4B,UAAU;IACtD"}
+{"version":3,"file":"output_parser.js","names":[],"sources":["../../../src/experimental/autogpt/output_parser.ts"],"sourcesContent":["import { BaseOutputParser } from \"@langchain/core/output_parsers\";\nimport { AutoGPTAction } from \"./schema.js\";\n\n/**\n * Utility function used to preprocess a string to be parsed as JSON.\n * It replaces single backslashes with double backslashes, while leaving\n * already escaped ones intact.\n * It also extracts the json code if it is inside a code block\n */\nexport function preprocessJsonInput(inputStr: string): string {\n  const correctedStr = inputStr.replace(\n    /(?<!\\\\)\\\\(?![\"\\\\/bfnrt]|u[0-9a-fA-F]{4})/g,\n    \"\\\\\\\\\"\n  );\n  const match = correctedStr.match(\n    /```(.*)(\\r\\n|\\r|\\n)(?<code>[\\w\\W\\n]+)(\\r\\n|\\r|\\n)```/\n  );\n  if (match?.groups?.code) {\n    return match.groups.code.trim();\n  } else {\n    return correctedStr;\n  }\n}\n\n/**\n * Class responsible for parsing the output of AutoGPT. It extends the\n * BaseOutputParser class.\n */\nexport class AutoGPTOutputParser extends BaseOutputParser<AutoGPTAction> {\n  lc_namespace = [\"langchain\", \"experimental\", \"autogpt\"];\n\n  /**\n   * Method not implemented in the class and will throw an error if called.\n   * It is likely meant to be overridden in subclasses to provide specific\n   * format instructions.\n   * @returns Throws an error.\n   */\n  getFormatInstructions(): string {\n    throw new Error(\"Method not implemented.\");\n  }\n\n  /**\n   * Asynchronous method that takes a string as input and attempts to parse\n   * it into an AutoGPTAction object. If the input string cannot be parsed\n   * directly, the method tries to preprocess the string using the\n   * preprocessJsonInput function and parse it again. If parsing fails\n   * again, it returns an AutoGPTAction object with an error message.\n   * @param text The string to be parsed.\n   * @returns A Promise that resolves to an AutoGPTAction object.\n   */\n  async parse(text: string): Promise<AutoGPTAction> {\n    let parsed: {\n      command: {\n        name: string;\n        args: Record<string, unknown>;\n      };\n    };\n    try {\n      parsed = JSON.parse(text);\n    } catch {\n      const preprocessedText = preprocessJsonInput(text);\n      try {\n        parsed = JSON.parse(preprocessedText);\n      } catch {\n        return {\n          name: \"ERROR\",\n          args: { error: `Could not parse invalid json: ${text}` },\n        };\n      }\n    }\n    try {\n      return {\n        name: parsed.command.name,\n        args: parsed.command.args,\n      };\n    } catch {\n      return {\n        name: \"ERROR\",\n        args: { error: `Incomplete command args: ${parsed}` },\n      };\n    }\n  }\n}\n"],"mappings":";;;;;;;;AASA,SAAgB,oBAAoB,UAA0B;CAC5D,MAAM,eAAe,SAAS,QAC5B,6CACA,MACF;CACA,MAAM,QAAQ,aAAa,MACzB,sDACF;CACA,IAAI,OAAO,QAAQ,MACjB,OAAO,MAAM,OAAO,KAAK,KAAK;MAE9B,OAAO;AAEX;;;;;AAMA,IAAa,sBAAb,cAAyC,iBAAgC;CACvE,eAAe;EAAC;EAAa;EAAgB;CAAS;;;;;;;CAQtD,wBAAgC;EAC9B,MAAM,IAAI,MAAM,yBAAyB;CAC3C;;;;;;;;;;CAWA,MAAM,MAAM,MAAsC;EAChD,IAAI;EAMJ,IAAI;GACF,SAAS,KAAK,MAAM,IAAI;EAC1B,QAAQ;GACN,MAAM,mBAAmB,oBAAoB,IAAI;GACjD,IAAI;IACF,SAAS,KAAK,MAAM,gBAAgB;GACtC,QAAQ;IACN,OAAO;KACL,MAAM;KACN,MAAM,EAAE,OAAO,iCAAiC,OAAO;IACzD;GACF;EACF;EACA,IAAI;GACF,OAAO;IACL,MAAM,OAAO,QAAQ;IACrB,MAAM,OAAO,QAAQ;GACvB;EACF,QAAQ;GACN,OAAO;IACL,MAAM;IACN,MAAM,EAAE,OAAO,4BAA4B,SAAS;GACtD;EACF;CACF;AACF"}

package/dist/experimental/autogpt/prompt.cjs

@@ -1,4 +1,3 @@
-require("../../_virtual/_rolldown/runtime.cjs");
 const require_prompt_generator = require("./prompt_generator.cjs");
 let _langchain_core_messages = require("@langchain/core/messages");
 let _langchain_core_prompts = require("@langchain/core/prompts");

package/dist/experimental/autogpt/prompt.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"prompt.cjs","names":["BaseChatPromptTemplate","getPrompt","SystemMessage","HumanMessage"],"sources":["../../../src/experimental/autogpt/prompt.ts"],"sourcesContent":["import type { VectorStoreRetrieverInterface } from \"@langchain/core/vectorstores\";\nimport {\n  BaseChatPromptTemplate,\n  SerializedBasePromptTemplate,\n} from \"@langchain/core/prompts\";\nimport {\n  BaseMessage,\n  HumanMessage,\n  SystemMessage,\n} from \"@langchain/core/messages\";\nimport { PartialValues } from \"@langchain/core/utils/types\";\nimport { getPrompt } from \"./prompt_generator.js\";\nimport { ObjectTool } from \"./schema.js\";\n\n/**\n * Interface for the input parameters of the AutoGPTPrompt class.\n */\nexport interface AutoGPTPromptInput {\n  aiName: string;\n  aiRole: string;\n  tools: ObjectTool[];\n  tokenCounter: (text: string) => Promise<number>;\n  sendTokenLimit?: number;\n}\n\n/**\n * Class used to generate prompts for the AutoGPT model. It takes into\n * account the AI's name, role, tools, token counter, and send token\n * limit. The class also handles the formatting of messages and the\n * construction of the full prompt.\n */\nexport class AutoGPTPrompt\n  extends BaseChatPromptTemplate\n  implements AutoGPTPromptInput\n{\n  aiName: string;\n\n  aiRole: string;\n\n  tools: ObjectTool[];\n\n  tokenCounter: (text: string) => Promise<number>;\n\n  sendTokenLimit: number;\n\n  constructor(fields: AutoGPTPromptInput) {\n    super({ inputVariables: [\"goals\", \"memory\", \"messages\", \"user_input\"] });\n    this.aiName = fields.aiName;\n    this.aiRole = fields.aiRole;\n    this.tools = fields.tools;\n    this.tokenCounter = fields.tokenCounter;\n    this.sendTokenLimit = fields.sendTokenLimit || 4196;\n  }\n\n  _getPromptType() {\n    return \"autogpt\" as const;\n  }\n\n  /**\n   * Constructs the full prompt based on the provided goals.\n   * @param goals An array of goals.\n   * @returns The full prompt as a string.\n   */\n  constructFullPrompt(goals: string[]): string {\n    const promptStart = `Your decisions must always be made independently\n            without seeking user assistance. Play to your strengths\n            as an LLM and pursue simple strategies with no legal complications.\n            If you have completed all your tasks,\n            make sure to use the \"finish\" command.`;\n\n    let fullPrompt = `You are ${this.aiName}, ${this.aiRole}\\n${promptStart}\\n\\nGOALS:\\n\\n`;\n    goals.forEach((goal, index) => {\n      fullPrompt += `${index + 1}. ${goal}\\n`;\n    });\n\n    fullPrompt += `\\n\\n${getPrompt(this.tools)}`;\n    return fullPrompt;\n  }\n\n  /**\n   * Formats the messages based on the provided parameters.\n   * @param goals An array of goals.\n   * @param memory A VectorStoreRetriever instance.\n   * @param messages An array of previous messages.\n   * @param user_input The user's input.\n   * @returns An array of formatted messages.\n   */\n  async formatMessages({\n    goals,\n    memory,\n    messages: previousMessages,\n    user_input,\n  }: {\n    goals: string[];\n    memory: VectorStoreRetrieverInterface;\n    messages: BaseMessage[];\n    user_input: string;\n  }) {\n    const basePrompt = new SystemMessage(this.constructFullPrompt(goals));\n    const timePrompt = new SystemMessage(\n      `The current time and date is ${new Date().toLocaleString(\"en-US\")}`\n    );\n    if (\n      typeof basePrompt.content !== \"string\" ||\n      typeof timePrompt.content !== \"string\"\n    ) {\n      throw new Error(\"Non-string message content is not supported.\");\n    }\n    const usedTokens =\n      (await this.tokenCounter(basePrompt.content)) +\n      (await this.tokenCounter(timePrompt.content));\n    const relevantDocs = await memory.invoke(\n      JSON.stringify(previousMessages.slice(-10))\n    );\n    const relevantMemory = relevantDocs.map(\n      (d: { pageContent: string }) => d.pageContent\n    );\n    let relevantMemoryTokens = await relevantMemory.reduce(\n      async (acc: Promise<number>, doc: string) =>\n        (await acc) + (await this.tokenCounter(doc)),\n      Promise.resolve(0)\n    );\n\n    while (usedTokens + relevantMemoryTokens > 2500) {\n      relevantMemory.pop();\n      relevantMemoryTokens = await relevantMemory.reduce(\n        async (acc: Promise<number>, doc: string) =>\n          (await acc) + (await this.tokenCounter(doc)),\n        Promise.resolve(0)\n      );\n    }\n\n    const contentFormat = `This reminds you of these events from your past:\\n${relevantMemory.join(\n      \"\\n\"\n    )}\\n\\n`;\n    const memoryMessage = new SystemMessage(contentFormat);\n    if (typeof memoryMessage.content !== \"string\") {\n      throw new Error(\"Non-string message content is not supported.\");\n    }\n    const usedTokensWithMemory =\n      usedTokens + (await this.tokenCounter(memoryMessage.content));\n    const historicalMessages: BaseMessage[] = [];\n\n    for (const message of previousMessages.slice(-10).reverse()) {\n      if (typeof message.content !== \"string\") {\n        throw new Error(\"Non-string message content is not supported.\");\n      }\n      const messageTokens = await this.tokenCounter(message.content);\n      if (usedTokensWithMemory + messageTokens > this.sendTokenLimit - 1000) {\n        break;\n      }\n      historicalMessages.unshift(message);\n    }\n\n    const inputMessage = new HumanMessage(user_input);\n    const messages: BaseMessage[] = [\n      basePrompt,\n      timePrompt,\n      memoryMessage,\n      ...historicalMessages,\n      inputMessage,\n    ];\n    return messages;\n  }\n\n  /**\n   * This method is not implemented in the AutoGPTPrompt class and will\n   * throw an error if called.\n   * @param _values Partial values.\n   * @returns Throws an error.\n   */\n  async partial(_values: PartialValues): Promise<BaseChatPromptTemplate> {\n    throw new Error(\"Method not implemented.\");\n  }\n\n  serialize(): SerializedBasePromptTemplate {\n    throw new Error(\"Method not implemented.\");\n  }\n}\n"],"mappings":";;;;;;;;;;;AA+BA,IAAa,gBAAb,cACUA,wBAAAA,uBAEV;CACE;CAEA;CAEA;CAEA;CAEA;CAEA,YAAY,QAA4B;AACtC,QAAM,EAAE,gBAAgB;GAAC;GAAS;GAAU;GAAY;GAAa,EAAE,CAAC;AACxE,OAAK,SAAS,OAAO;AACrB,OAAK,SAAS,OAAO;AACrB,OAAK,QAAQ,OAAO;AACpB,OAAK,eAAe,OAAO;AAC3B,OAAK,iBAAiB,OAAO,kBAAkB;;CAGjD,iBAAiB;AACf,SAAO;;;;;;;CAQT,oBAAoB,OAAyB;EAO3C,IAAI,aAAa,WAAW,KAAK,OAAO,IAAI,KAAK,OAAO;;;;;AACxD,QAAM,SAAS,MAAM,UAAU;AAC7B,iBAAc,GAAG,QAAQ,EAAE,IAAI,KAAK;IACpC;AAEF,gBAAc,OAAOC,yBAAAA,UAAU,KAAK,MAAM;AAC1C,SAAO;;;;;;;;;;CAWT,MAAM,eAAe,EACnB,OACA,QACA,UAAU,kBACV,cAMC;EACD,MAAM,aAAa,IAAIC,yBAAAA,cAAc,KAAK,oBAAoB,MAAM,CAAC;EACrE,MAAM,aAAa,IAAIA,yBAAAA,cACrB,iDAAgC,IAAI,MAAM,EAAC,eAAe,QAAQ,GACnE;AACD,MACE,OAAO,WAAW,YAAY,YAC9B,OAAO,WAAW,YAAY,SAE9B,OAAM,IAAI,MAAM,+CAA+C;EAEjE,MAAM,aACH,MAAM,KAAK,aAAa,WAAW,QAAQ,GAC3C,MAAM,KAAK,aAAa,WAAW,QAAQ;EAI9C,MAAM,kBAAiB,MAHI,OAAO,OAChC,KAAK,UAAU,iBAAiB,MAAM,IAAI,CAAC,CAC5C,EACmC,KACjC,MAA+B,EAAE,YACnC;EACD,IAAI,uBAAuB,MAAM,eAAe,OAC9C,OAAO,KAAsB,QAC1B,MAAM,MAAQ,MAAM,KAAK,aAAa,IAAI,EAC7C,QAAQ,QAAQ,EAAE,CACnB;AAED,SAAO,aAAa,uBAAuB,MAAM;AAC/C,kBAAe,KAAK;AACpB,0BAAuB,MAAM,eAAe,OAC1C,OAAO,KAAsB,QAC1B,MAAM,MAAQ,MAAM,KAAK,aAAa,IAAI,EAC7C,QAAQ,QAAQ,EAAE,CACnB;;EAMH,MAAM,gBAAgB,IAAIA,yBAAAA,cAAc,qDAHmC,eAAe,KACxF,KACD,CAAC,MACoD;AACtD,MAAI,OAAO,cAAc,YAAY,SACnC,OAAM,IAAI,MAAM,+CAA+C;EAEjE,MAAM,uBACJ,aAAc,MAAM,KAAK,aAAa,cAAc,QAAQ;EAC9D,MAAM,qBAAoC,EAAE;AAE5C,OAAK,MAAM,WAAW,iBAAiB,MAAM,IAAI,CAAC,SAAS,EAAE;AAC3D,OAAI,OAAO,QAAQ,YAAY,SAC7B,OAAM,IAAI,MAAM,+CAA+C;AAGjE,OAAI,uBAAuB,MADC,KAAK,aAAa,QAAQ,QAAQ,GACnB,KAAK,iBAAiB,IAC/D;AAEF,sBAAmB,QAAQ,QAAQ;;EAGrC,MAAM,eAAe,IAAIC,yBAAAA,aAAa,WAAW;AAQjD,SAAO;GANL;GACA;GACA;GACA,GAAG;GACH;GAEa;;;;;;;;CASjB,MAAM,QAAQ,SAAyD;AACrE,QAAM,IAAI,MAAM,0BAA0B;;CAG5C,YAA0C;AACxC,QAAM,IAAI,MAAM,0BAA0B"}
+{"version":3,"file":"prompt.cjs","names":["BaseChatPromptTemplate","getPrompt","SystemMessage","HumanMessage"],"sources":["../../../src/experimental/autogpt/prompt.ts"],"sourcesContent":["import type { VectorStoreRetrieverInterface } from \"@langchain/core/vectorstores\";\nimport {\n  BaseChatPromptTemplate,\n  SerializedBasePromptTemplate,\n} from \"@langchain/core/prompts\";\nimport {\n  BaseMessage,\n  HumanMessage,\n  SystemMessage,\n} from \"@langchain/core/messages\";\nimport { PartialValues } from \"@langchain/core/utils/types\";\nimport { getPrompt } from \"./prompt_generator.js\";\nimport { ObjectTool } from \"./schema.js\";\n\n/**\n * Interface for the input parameters of the AutoGPTPrompt class.\n */\nexport interface AutoGPTPromptInput {\n  aiName: string;\n  aiRole: string;\n  tools: ObjectTool[];\n  tokenCounter: (text: string) => Promise<number>;\n  sendTokenLimit?: number;\n}\n\n/**\n * Class used to generate prompts for the AutoGPT model. It takes into\n * account the AI's name, role, tools, token counter, and send token\n * limit. The class also handles the formatting of messages and the\n * construction of the full prompt.\n */\nexport class AutoGPTPrompt\n  extends BaseChatPromptTemplate\n  implements AutoGPTPromptInput\n{\n  aiName: string;\n\n  aiRole: string;\n\n  tools: ObjectTool[];\n\n  tokenCounter: (text: string) => Promise<number>;\n\n  sendTokenLimit: number;\n\n  constructor(fields: AutoGPTPromptInput) {\n    super({ inputVariables: [\"goals\", \"memory\", \"messages\", \"user_input\"] });\n    this.aiName = fields.aiName;\n    this.aiRole = fields.aiRole;\n    this.tools = fields.tools;\n    this.tokenCounter = fields.tokenCounter;\n    this.sendTokenLimit = fields.sendTokenLimit || 4196;\n  }\n\n  _getPromptType() {\n    return \"autogpt\" as const;\n  }\n\n  /**\n   * Constructs the full prompt based on the provided goals.\n   * @param goals An array of goals.\n   * @returns The full prompt as a string.\n   */\n  constructFullPrompt(goals: string[]): string {\n    const promptStart = `Your decisions must always be made independently\n            without seeking user assistance. Play to your strengths\n            as an LLM and pursue simple strategies with no legal complications.\n            If you have completed all your tasks,\n            make sure to use the \"finish\" command.`;\n\n    let fullPrompt = `You are ${this.aiName}, ${this.aiRole}\\n${promptStart}\\n\\nGOALS:\\n\\n`;\n    goals.forEach((goal, index) => {\n      fullPrompt += `${index + 1}. ${goal}\\n`;\n    });\n\n    fullPrompt += `\\n\\n${getPrompt(this.tools)}`;\n    return fullPrompt;\n  }\n\n  /**\n   * Formats the messages based on the provided parameters.\n   * @param goals An array of goals.\n   * @param memory A VectorStoreRetriever instance.\n   * @param messages An array of previous messages.\n   * @param user_input The user's input.\n   * @returns An array of formatted messages.\n   */\n  async formatMessages({\n    goals,\n    memory,\n    messages: previousMessages,\n    user_input,\n  }: {\n    goals: string[];\n    memory: VectorStoreRetrieverInterface;\n    messages: BaseMessage[];\n    user_input: string;\n  }) {\n    const basePrompt = new SystemMessage(this.constructFullPrompt(goals));\n    const timePrompt = new SystemMessage(\n      `The current time and date is ${new Date().toLocaleString(\"en-US\")}`\n    );\n    if (\n      typeof basePrompt.content !== \"string\" ||\n      typeof timePrompt.content !== \"string\"\n    ) {\n      throw new Error(\"Non-string message content is not supported.\");\n    }\n    const usedTokens =\n      (await this.tokenCounter(basePrompt.content)) +\n      (await this.tokenCounter(timePrompt.content));\n    const relevantDocs = await memory.invoke(\n      JSON.stringify(previousMessages.slice(-10))\n    );\n    const relevantMemory = relevantDocs.map(\n      (d: { pageContent: string }) => d.pageContent\n    );\n    let relevantMemoryTokens = await relevantMemory.reduce(\n      async (acc: Promise<number>, doc: string) =>\n        (await acc) + (await this.tokenCounter(doc)),\n      Promise.resolve(0)\n    );\n\n    while (usedTokens + relevantMemoryTokens > 2500) {\n      relevantMemory.pop();\n      relevantMemoryTokens = await relevantMemory.reduce(\n        async (acc: Promise<number>, doc: string) =>\n          (await acc) + (await this.tokenCounter(doc)),\n        Promise.resolve(0)\n      );\n    }\n\n    const contentFormat = `This reminds you of these events from your past:\\n${relevantMemory.join(\n      \"\\n\"\n    )}\\n\\n`;\n    const memoryMessage = new SystemMessage(contentFormat);\n    if (typeof memoryMessage.content !== \"string\") {\n      throw new Error(\"Non-string message content is not supported.\");\n    }\n    const usedTokensWithMemory =\n      usedTokens + (await this.tokenCounter(memoryMessage.content));\n    const historicalMessages: BaseMessage[] = [];\n\n    for (const message of previousMessages.slice(-10).reverse()) {\n      if (typeof message.content !== \"string\") {\n        throw new Error(\"Non-string message content is not supported.\");\n      }\n      const messageTokens = await this.tokenCounter(message.content);\n      if (usedTokensWithMemory + messageTokens > this.sendTokenLimit - 1000) {\n        break;\n      }\n      historicalMessages.unshift(message);\n    }\n\n    const inputMessage = new HumanMessage(user_input);\n    const messages: BaseMessage[] = [\n      basePrompt,\n      timePrompt,\n      memoryMessage,\n      ...historicalMessages,\n      inputMessage,\n    ];\n    return messages;\n  }\n\n  /**\n   * This method is not implemented in the AutoGPTPrompt class and will\n   * throw an error if called.\n   * @param _values Partial values.\n   * @returns Throws an error.\n   */\n  async partial(_values: PartialValues): Promise<BaseChatPromptTemplate> {\n    throw new Error(\"Method not implemented.\");\n  }\n\n  serialize(): SerializedBasePromptTemplate {\n    throw new Error(\"Method not implemented.\");\n  }\n}\n"],"mappings":";;;;;;;;;;AA+BA,IAAa,gBAAb,cACUA,wBAAAA,uBAEV;CACE;CAEA;CAEA;CAEA;CAEA;CAEA,YAAY,QAA4B;EACtC,MAAM,EAAE,gBAAgB;GAAC;GAAS;GAAU;GAAY;EAAY,EAAE,CAAC;EACvE,KAAK,SAAS,OAAO;EACrB,KAAK,SAAS,OAAO;EACrB,KAAK,QAAQ,OAAO;EACpB,KAAK,eAAe,OAAO;EAC3B,KAAK,iBAAiB,OAAO,kBAAkB;CACjD;CAEA,iBAAiB;EACf,OAAO;CACT;;;;;;CAOA,oBAAoB,OAAyB;EAO3C,IAAI,aAAa,WAAW,KAAK,OAAO,IAAI,KAAK,OAAO;;;;;EACxD,MAAM,SAAS,MAAM,UAAU;GAC7B,cAAc,GAAG,QAAQ,EAAE,IAAI,KAAK;EACtC,CAAC;EAED,cAAc,OAAOC,yBAAAA,UAAU,KAAK,KAAK;EACzC,OAAO;CACT;;;;;;;;;CAUA,MAAM,eAAe,EACnB,OACA,QACA,UAAU,kBACV,cAMC;EACD,MAAM,aAAa,IAAIC,yBAAAA,cAAc,KAAK,oBAAoB,KAAK,CAAC;EACpE,MAAM,aAAa,IAAIA,yBAAAA,cACrB,iDAAgC,IAAI,KAAK,EAAA,CAAE,eAAe,OAAO,GACnE;EACA,IACE,OAAO,WAAW,YAAY,YAC9B,OAAO,WAAW,YAAY,UAE9B,MAAM,IAAI,MAAM,8CAA8C;EAEhE,MAAM,aACH,MAAM,KAAK,aAAa,WAAW,OAAO,IAC1C,MAAM,KAAK,aAAa,WAAW,OAAO;EAI7C,MAAM,kBAAiB,MAHI,OAAO,OAChC,KAAK,UAAU,iBAAiB,MAAM,GAAG,CAAC,CAC5C,EAAA,CACoC,KACjC,MAA+B,EAAE,WACpC;EACA,IAAI,uBAAuB,MAAM,eAAe,OAC9C,OAAO,KAAsB,QAC1B,MAAM,MAAQ,MAAM,KAAK,aAAa,GAAG,GAC5C,QAAQ,QAAQ,CAAC,CACnB;EAEA,OAAO,aAAa,uBAAuB,MAAM;GAC/C,eAAe,IAAI;GACnB,uBAAuB,MAAM,eAAe,OAC1C,OAAO,KAAsB,QAC1B,MAAM,MAAQ,MAAM,KAAK,aAAa,GAAG,GAC5C,QAAQ,QAAQ,CAAC,CACnB;EACF;EAKA,MAAM,gBAAgB,IAAIA,yBAAAA,cAAc,qDAHmC,eAAe,KACxF,IACF,EAAE,KACmD;EACrD,IAAI,OAAO,cAAc,YAAY,UACnC,MAAM,IAAI,MAAM,8CAA8C;EAEhE,MAAM,uBACJ,aAAc,MAAM,KAAK,aAAa,cAAc,OAAO;EAC7D,MAAM,qBAAoC,CAAC;EAE3C,KAAK,MAAM,WAAW,iBAAiB,MAAM,GAAG,CAAC,CAAC,QAAQ,GAAG;GAC3D,IAAI,OAAO,QAAQ,YAAY,UAC7B,MAAM,IAAI,MAAM,8CAA8C;GAGhE,IAAI,uBAAuB,MADC,KAAK,aAAa,QAAQ,OAAO,IAClB,KAAK,iBAAiB,KAC/D;GAEF,mBAAmB,QAAQ,OAAO;EACpC;EAEA,MAAM,eAAe,IAAIC,yBAAAA,aAAa,UAAU;EAQhD,OAAO;GANL;GACA;GACA;GACA,GAAG;GACH;EAEY;CAChB;;;;;;;CAQA,MAAM,QAAQ,SAAyD;EACrE,MAAM,IAAI,MAAM,yBAAyB;CAC3C;CAEA,YAA0C;EACxC,MAAM,IAAI,MAAM,yBAAyB;CAC3C;AACF"}

package/dist/experimental/autogpt/prompt.d.cts

@@ -1,10 +1,8 @@
 import { ObjectTool } from "./schema.cjs";
-import * as _$_langchain_core_messages0 from "@langchain/core/messages";
 import { BaseMessage } from "@langchain/core/messages";
 import { PartialValues } from "@langchain/core/utils/types";
 import { BaseChatPromptTemplate, SerializedBasePromptTemplate } from "@langchain/core/prompts";
 import { VectorStoreRetrieverInterface } from "@langchain/core/vectorstores";
-
 //#region src/experimental/autogpt/prompt.d.ts
 /**
  * Interface for the input parameters of the AutoGPTPrompt class.
@@ -44,17 +42,12 @@ declare class AutoGPTPrompt extends BaseChatPromptTemplate implements AutoGPTPro
    * @param user_input The user's input.
    * @returns An array of formatted messages.
    */
-  formatMessages({
-    goals,
-    memory,
-    messages: previousMessages,
-    user_input
-  }: {
+  formatMessages({ goals, memory, messages: previousMessages, user_input }: {
     goals: string[];
     memory: VectorStoreRetrieverInterface;
     messages: BaseMessage[];
     user_input: string;
-  }): Promise<BaseMessage<_$_langchain_core_messages0.MessageStructure<_$_langchain_core_messages0.MessageToolSet>, _$_langchain_core_messages0.MessageType>[]>;
+  }): Promise<BaseMessage<import("@langchain/core/messages").MessageStructure<import("@langchain/core/messages").MessageToolSet>, import("@langchain/core/messages").MessageType>[]>;
   /**
    * This method is not implemented in the AutoGPTPrompt class and will
    * throw an error if called.

package/dist/experimental/autogpt/prompt.d.cts.map

@@ -1 +1 @@
-{"version":3,"file":"prompt.d.cts","names":[],"sources":["../../../src/experimental/autogpt/prompt.ts"],"mappings":";;;;;;;;;;;UAiBiB,kBAAA;EACf,MAAA;EACA,MAAA;EACA,KAAA,EAAO,UAAA;EACP,YAAA,GAAe,IAAA,aAAiB,OAAA;EAChC,cAAA;AAAA;;;;;;;cASW,aAAA,SACH,sBAAA,YACG,kBAAA;EAEX,MAAA;EAEA,MAAA;EAEA,KAAA,EAAO,UAAA;EAEP,YAAA,GAAe,IAAA,aAAiB,OAAA;EAEhC,cAAA;EAEA,WAAA,CAAY,MAAA,EAAQ,kBAAA;EASpB,cAAA,CAAA;EAmCE;;;;;EA1BF,mBAAA,CAAoB,KAAA;EAkCnB;;;;;;;;EAVK,cAAA,CAAA;IACJ,KAAA;IACA,MAAA;IACA,QAAA,EAAU,gBAAA;IACV;EAAA;IAEA,KAAA;IACA,MAAA,EAAQ,6BAAA;IACR,QAAA,EAAU,WAAA;IACV,UAAA;EAAA,IACD,OAAA,CAAA,WAAA,CAAA,2BAAA,CAAA,gBAAA,CAFsB,2BAAA,CAEtB,cAAA,GAAA,2BAAA,CAAA,WAAA;EA9DD;;;;;;EAwIM,OAAA,CAAQ,OAAA,EAAS,aAAA,GAAgB,OAAA,CAAQ,sBAAA;EAI/C,SAAA,CAAA,GAAa,4BAAA;AAAA"}
+{"version":3,"file":"prompt.d.cts","names":[],"sources":["../../../src/experimental/autogpt/prompt.ts"],"mappings":";;;;;;;;;UAiBiB;EACf;EACA;EACA,OAAO;EACP,eAAe,iBAAiB;EAChC;;;;;;;;cASW,sBACH,kCACG;EAEX;EAEA;EAEA,OAAO;EAEP,eAAe,iBAAiB;EAEhC;EAEA,YAAY,QAAQ;EASpB;;;;;;EASA,oBAAoB;;;;;;;;;EAwBd,iBACJ,OACA,QACA,UAAU,kBACV;IAEA;IACA,QAAQ;IACR,UAAU;IACV;MACD,QAAA,+CAAA,oDAAA,oDAAA;;;;;;;EA0EK,QAAQ,SAAS,gBAAgB,QAAQ;EAI/C,aAAa"}