alchemy 2.0.0-beta.37 → 2.0.0-beta.39

package/src/Agent/llm/error.ts

@@ -1,330 +0,0 @@
-import * as AISDKProvider from "@ai-sdk/provider";
-import { DownloadError as AISDKDownloadError } from "ai";
-import * as Data from "effect/Data";
-
-/**
- * Base AI SDK error - wraps generic AI SDK errors
- */
-export class AISDKError extends Data.TaggedError("AISDKError")<{
-  readonly message: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * API call error - wraps errors from API calls to AI providers
- */
-export class APICallError extends Data.TaggedError("APICallError")<{
-  readonly message: string;
-  readonly url: string;
-  readonly requestBodyValues: unknown;
-  readonly statusCode?: number;
-  readonly responseHeaders?: Record<string, string>;
-  readonly responseBody?: string;
-  readonly isRetryable: boolean;
-  readonly data?: unknown;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Empty response body error - thrown when the API returns an empty response
- */
-export class EmptyResponseBodyError extends Data.TaggedError(
-  "EmptyResponseBodyError",
-)<{
-  readonly message: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Invalid argument error - thrown when a function argument is invalid
- */
-export class InvalidArgumentError extends Data.TaggedError(
-  "InvalidArgumentError",
-)<{
-  readonly message: string;
-  readonly argument: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Invalid prompt error - thrown when a prompt cannot be processed
- */
-export class InvalidPromptError extends Data.TaggedError("InvalidPromptError")<{
-  readonly message: string;
-  readonly prompt: unknown;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Invalid response data error - thrown when response data cannot be parsed
- */
-export class InvalidResponseDataError extends Data.TaggedError(
-  "InvalidResponseDataError",
-)<{
-  readonly message: string;
-  readonly data: unknown;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * JSON parse error - thrown when JSON parsing fails
- */
-export class JSONParseError extends Data.TaggedError("JSONParseError")<{
-  readonly message: string;
-  readonly text: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Load API key error - thrown when API key cannot be loaded
- */
-export class LoadAPIKeyError extends Data.TaggedError("LoadAPIKeyError")<{
-  readonly message: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Load setting error - thrown when a setting cannot be loaded
- */
-export class LoadSettingError extends Data.TaggedError("LoadSettingError")<{
-  readonly message: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * No content generated error - thrown when the AI provider fails to generate content
- */
-export class NoContentGeneratedError extends Data.TaggedError(
-  "NoContentGeneratedError",
-)<{
-  readonly message: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * No such model error - thrown when a model cannot be found
- */
-export class NoSuchModelError extends Data.TaggedError("NoSuchModelError")<{
-  readonly message: string;
-  readonly modelId: string;
-  readonly modelType: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Too many embedding values error - thrown when too many values are passed to an embedding call
- */
-export class TooManyEmbeddingValuesForCallError extends Data.TaggedError(
-  "TooManyEmbeddingValuesForCallError",
-)<{
-  readonly message: string;
-  readonly provider: string;
-  readonly modelId: string;
-  readonly maxEmbeddingsPerCall: number;
-  readonly values: Array<unknown>;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Type validation error - thrown when type validation fails
- */
-export class TypeValidationError extends Data.TaggedError(
-  "TypeValidationError",
-)<{
-  readonly message: string;
-  readonly value: unknown;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Unsupported functionality error - thrown when a feature is not supported
- */
-export class UnsupportedFunctionalityError extends Data.TaggedError(
-  "UnsupportedFunctionalityError",
-)<{
-  readonly message: string;
-  readonly functionality: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Download error - thrown when a file download fails
- */
-export class DownloadError extends Data.TaggedError("DownloadError")<{
-  readonly message: string;
-  readonly url: string;
-  readonly statusCode?: number;
-  readonly statusText?: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Unknown error - fallback for unrecognized errors
- */
-export class UnknownAISDKError extends Data.TaggedError("UnknownAISDKError")<{
-  readonly message: string;
-  readonly cause?: unknown;
-}> {}
-
-/**
- * Union type of all AI SDK errors
- */
-export type LLMError =
-  | AISDKError
-  | APICallError
-  | EmptyResponseBodyError
-  | InvalidArgumentError
-  | InvalidPromptError
-  | InvalidResponseDataError
-  | JSONParseError
-  | LoadAPIKeyError
-  | LoadSettingError
-  | NoContentGeneratedError
-  | NoSuchModelError
-  | TooManyEmbeddingValuesForCallError
-  | TypeValidationError
-  | UnsupportedFunctionalityError
-  | DownloadError
-  | UnknownAISDKError;
-
-/**
- * Converts any error to a typed LLMError
- */
-export const fromAnyError = (e: unknown): LLMError => {
-  if (AISDKDownloadError.isInstance(e)) {
-    return new DownloadError({
-      message: e.message,
-      url: e.url,
-      statusCode: e.statusCode,
-      statusText: e.statusText,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.APICallError.isInstance(e)) {
-    return new APICallError({
-      message: e.message,
-      url: e.url,
-      requestBodyValues: e.requestBodyValues,
-      statusCode: e.statusCode,
-      responseHeaders: e.responseHeaders,
-      responseBody: e.responseBody,
-      isRetryable: e.isRetryable,
-      data: e.data,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.EmptyResponseBodyError.isInstance(e)) {
-    return new EmptyResponseBodyError({
-      message: e.message,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.InvalidArgumentError.isInstance(e)) {
-    return new InvalidArgumentError({
-      message: e.message,
-      argument: e.argument,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.InvalidPromptError.isInstance(e)) {
-    return new InvalidPromptError({
-      message: e.message,
-      prompt: e.prompt,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.InvalidResponseDataError.isInstance(e)) {
-    return new InvalidResponseDataError({
-      message: e.message,
-      data: e.data,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.JSONParseError.isInstance(e)) {
-    return new JSONParseError({
-      message: e.message,
-      text: e.text,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.LoadAPIKeyError.isInstance(e)) {
-    return new LoadAPIKeyError({
-      message: e.message,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.LoadSettingError.isInstance(e)) {
-    return new LoadSettingError({
-      message: e.message,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.NoContentGeneratedError.isInstance(e)) {
-    return new NoContentGeneratedError({
-      message: e.message,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.NoSuchModelError.isInstance(e)) {
-    return new NoSuchModelError({
-      message: e.message,
-      modelId: e.modelId,
-      modelType: e.modelType,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.TooManyEmbeddingValuesForCallError.isInstance(e)) {
-    return new TooManyEmbeddingValuesForCallError({
-      message: e.message,
-      provider: e.provider,
-      modelId: e.modelId,
-      maxEmbeddingsPerCall: e.maxEmbeddingsPerCall,
-      values: e.values,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.TypeValidationError.isInstance(e)) {
-    return new TypeValidationError({
-      message: e.message,
-      value: e.value,
-      cause: e,
-    });
-  }
-
-  if (AISDKProvider.UnsupportedFunctionalityError.isInstance(e)) {
-    return new UnsupportedFunctionalityError({
-      message: e.message,
-      functionality: e.functionality,
-      cause: e,
-    });
-  }
-
-  // Generic AISDKError check (should be last since it's the base class)
-  if (AISDKProvider.AISDKError.isInstance(e)) {
-    return new AISDKError({
-      message: e.message,
-      cause: e,
-    });
-  }
-
-  // Fallback for unknown errors
-  const message =
-    e instanceof Error
-      ? e.message
-      : typeof e === "string"
-        ? e
-        : "Unknown error";
-  return new UnknownAISDKError({ message, cause: e });
-};

package/src/Agent/llm/index.ts

@@ -1,4 +0,0 @@
-export * from "./error.ts";
-export * from "./llm.ts";
-export * from "./stream-collectors.ts";
-export * from "./stream-text-part.ts";

package/src/Agent/llm/llm.ts

@@ -1,131 +0,0 @@
-import {
-  createAnthropic,
-  type AnthropicProviderOptions,
-} from "@ai-sdk/anthropic";
-import {
-  createOpenAI,
-  type OpenAIResponsesProviderOptions,
-} from "@ai-sdk/openai";
-import { jsonSchema, tool, ToolLoopAgent, type ModelMessage } from "ai";
-import * as Context from "effect/Context";
-import * as Effect from "effect/Effect";
-import * as Layer from "effect/Layer";
-import * as Redacted from "effect/Redacted";
-import * as S from "effect/Schema";
-import * as Stream from "effect/Stream";
-import type { IsUnknown } from "../../Util/unknown.ts";
-import type { Tool } from "../tool/tool.ts";
-import { fromAnyError, type LLMError } from "./error.ts";
-import type { StreamTextPart } from "./stream-text-part.ts";
-
-export class LLM extends Context.Service<LLM, LLMService>()("LLM") {}
-
-export type StreamTextOptions<T extends Tool = Tool> = {
-  messages: ModelMessage[];
-  model?: string;
-  system: string;
-  tools: T[];
-};
-
-export interface LLMService {
-  stream: <Tools extends Tool>(
-    input: StreamTextOptions<Tools>,
-  ) => Stream.Stream<
-    StreamTextPart,
-    IsUnknown<Tool.Error<Tools>> extends true
-      ? LLMError
-      : LLMError | Tool.Error<Tools>,
-    Exclude<Tool.Context<Tools>, unknown>
-  >;
-}
-
-export interface LLMConfig {
-  anthropic?: {
-    apiKey: Redacted.Redacted<string>;
-  };
-  openai?: {
-    apiKey: Redacted.Redacted<string>;
-  };
-}
-
-export const llm = (config: LLMConfig) =>
-  Layer.sync(LLM, () => {
-    const anthropic = config.anthropic?.apiKey
-      ? createAnthropic({
-          apiKey: Redacted.value(config.anthropic?.apiKey),
-        })
-      : undefined;
-    const openai = config.openai?.apiKey
-      ? createOpenAI({ apiKey: Redacted.value(config.openai?.apiKey) })
-      : undefined;
-
-    return {
-      stream: <T extends Tool>(input: StreamTextOptions<T>) => {
-        const modelId = input.model ?? "anthropic/claude-opus-4.5";
-        const modelProvider = modelId.split("/")[0];
-        const model = modelId.includes("anthropic")
-          ? anthropic?.(modelId)
-          : openai?.(modelId);
-
-        if (!model) {
-          return Stream.die(`No model found for ${modelId}`);
-        }
-
-        return Stream.fromEffect(
-          Effect.gen(function* () {
-            // we need to get the caller's context so we can construct async functions
-            const context = yield* Effect.context<never>();
-            const tools: any = {};
-            for (const t of input.tools) {
-              const decode = S.decodeEffect(t.schema);
-              tools[t.props.alias?.(modelId) ?? t.id] = tool({
-                inputSchema: jsonSchema(S.toJsonSchemaDocument(t.schema)),
-                description: "TODO",
-                execute: (params) =>
-                  Effect.runPromise(
-                    decode(params as any).pipe(
-                      Effect.flatMap(t.handle),
-                      // provide the caller's context so the tool can access it
-                      Effect.provide(context),
-                    ) as Effect.Effect<any>,
-                  ),
-              });
-            }
-            return tools;
-          }).pipe(
-            Effect.flatMap((tools) =>
-              Effect.promise((abortSignal) =>
-                new ToolLoopAgent({
-                  model,
-                  instructions: input.system,
-                  // never stop, build the world!!!
-                  stopWhen: () => false,
-                  tools,
-                  providerOptions:
-                    modelProvider === "anthropic"
-                      ? {
-                          anthropic: {
-                            effort: "high",
-                            thinking: { type: "enabled", budgetTokens: 12000 },
-                          } satisfies AnthropicProviderOptions,
-                        }
-                      : {
-                          openai: {
-                            reasoningEffort: "high",
-                          } satisfies OpenAIResponsesProviderOptions,
-                        },
-                })
-                  .stream({
-                    messages: input.messages,
-                    abortSignal,
-                  })
-                  .then((s) => s.fullStream),
-              ),
-            ),
-          ),
-        ).pipe(
-          Stream.flatMap((it) => Stream.fromAsyncIterable(it, fromAnyError)),
-        );
-      },
-    };
-  });

package/src/Agent/llm/stream-collectors.ts

@@ -1,58 +0,0 @@
-import * as Sink from "effect/Sink";
-import type { StreamTextPart } from "./stream-text-part.ts";
-
-// Extract specific part types from the StreamTextPart union
-export type TextDeltaPart = Extract<StreamTextPart, { type: "text-delta" }>;
-export type TextStartPart = Extract<StreamTextPart, { type: "text-start" }>;
-export type TextEndPart = Extract<StreamTextPart, { type: "text-end" }>;
-export type ReasoningDeltaPart = Extract<
-  StreamTextPart,
-  { type: "reasoning-delta" }
->;
-export type ReasoningStartPart = Extract<
-  StreamTextPart,
-  { type: "reasoning-start" }
->;
-export type ReasoningEndPart = Extract<
-  StreamTextPart,
-  { type: "reasoning-end" }
->;
-export type ToolCallPart = Extract<StreamTextPart, { type: "tool-call" }>;
-export type ToolResultPart = Extract<StreamTextPart, { type: "tool-result" }>;
-export type ToolErrorPart = Extract<StreamTextPart, { type: "tool-error" }>;
-
-/**
- * Creates a sink that collects the text from the last complete text message.
- * A complete message is one that has received a text-end event.
- * Returns undefined if no complete text messages are present.
- *
- * @example
- * ```typescript
- * const lastText = yield* llm.stream({ ... }).pipe(
- *   Stream.run(collectLastText())
- * );
- * if (lastText) {
- *   yield* Console.log(`Last message: ${lastText}`);
- * }
- * ```
- */
-export const lastMessageText: Sink.Sink<string | undefined, StreamTextPart> =
-  Sink.reduce(
-    () => ({
-      currentId: undefined as string | undefined,
-      text: "",
-      lastComplete: undefined as string | undefined,
-    }),
-    (acc, part: StreamTextPart) => {
-      if (part.type === "text-start") {
-        return { currentId: part.id, text: "", lastComplete: acc.lastComplete };
-      }
-      if (part.type === "text-delta" && acc.currentId === part.id) {
-        return { ...acc, text: acc.text + part.text };
-      }
-      if (part.type === "text-end" && acc.currentId === part.id) {
-        return { ...acc, lastComplete: acc.text };
-      }
-      return acc;
-    },
-  ).pipe(Sink.map((acc) => acc.lastComplete));

package/src/Agent/llm/stream-text-part.ts

@@ -1,7 +0,0 @@
-import type { TextStreamPart } from "ai";
-import * as S from "effect/Schema";
-
-export type StreamTextPart = TextStreamPart<any>;
-
-// TODO(sam): build out the schema for this
-export const StreamTextPart = S.suspend((): S.Schema<StreamTextPart> => S.Any);