@tool-bridge/server 0.21.1 → 0.22.1

package/dist/cohere-XCJ5AJ6B.js

@@ -1,185 +0,0 @@
-import {
-  createAuthedClient
-} from "./chunk-XYBU3WZH.js";
-import "./chunk-E6JCVKW6.js";
-import {
-  asJsonObject,
-  trimmedText
-} from "./chunk-JXHMDBAY.js";
-import {
-  createProviderPlugin,
-  upstreamError
-} from "./chunk-ATK45BXV.js";
-import "./chunk-U7B6DS2O.js";
-import {
-  external_exports
-} from "./chunk-UZM2RXZH.js";
-import "./chunk-U67V476Y.js";
-
-// ../plugins/src/cohere/api.ts
-var SERVICE = "cohere";
-function errorMessage(payload, status) {
-  const fallback = trimmedText(payload) ?? `cohere request failed with ${status}`;
-  const root = asJsonObject(payload);
-  return trimmedText(asJsonObject(root?.error)?.message) ?? trimmedText(root?.message) ?? fallback;
-}
-var http = createAuthedClient({
-  baseUrl: "https://api.cohere.com/",
-  service: SERVICE,
-  auth: { kind: "bearer" },
-  headers: { accept: "application/json" },
-  // 498 是 Cohere 表达"token 无效"的私有码,改判 401 让它落 permission_denied。
-  mapError: ({ data: payload, status }) => upstreamError(
-    status === 498 ? 401 : status,
-    errorMessage(payload, status)
-  )
-});
-async function post(ctx, path, body) {
-  const { data } = await http.request(ctx, {
-    path,
-    method: "POST",
-    json: body,
-    invalidJsonMessage: "cohere returned an invalid JSON response"
-  });
-  const result = asJsonObject(data);
-  if (result === void 0) throw upstreamError(502, "cohere returned an invalid JSON response");
-  return result;
-}
-function chat(input, ctx) {
-  return post(ctx, "v2/chat", input);
-}
-function embedTexts(input, ctx) {
-  return post(ctx, "v2/embed", input);
-}
-function rerankDocuments(input, ctx) {
-  return post(ctx, "v2/rerank", input);
-}
-
-// ../plugins/src/cohere/schema.ts
-var chatInput = external_exports.strictObject({
-  model: external_exports.string().min(1).describe("The name of a compatible Cohere chat model."),
-  messages: external_exports.array(external_exports.strictObject({
-    role: external_exports.enum(["system", "user", "assistant", "tool"]).describe("The author role for the message."),
-    content: external_exports.union([external_exports.string().describe("Plain text message content."), external_exports.array(external_exports.looseObject({
-      type: external_exports.string().describe("The content block type, such as text or image_url.")
-    }).describe("A Cohere chat message content block.")).min(1).describe("Structured content blocks for the message.")]).describe("The content of a Cohere chat message.").optional(),
-    tool_calls: external_exports.array(external_exports.looseObject({}).describe("A JSON object accepted or returned by the Cohere API.")).describe("Tool calls made by the assistant message.").optional(),
-    tool_call_id: external_exports.string().describe("The identifier of the tool call this tool message responds to.").optional()
-  }).describe("A message in the conversation sent to the Cohere Chat API.")).min(1).describe("The ordered conversation messages to send to Cohere."),
-  tools: external_exports.array(external_exports.looseObject({}).describe("A JSON object accepted or returned by the Cohere API.")).describe("Function tools available to the model.").optional(),
-  strict_tools: external_exports.boolean().describe("Whether tool calls must strictly follow the tool definition.").optional(),
-  documents: external_exports.array(external_exports.union([external_exports.string().describe("Document text."), external_exports.looseObject({}).describe("A JSON object accepted or returned by the Cohere API.")]).describe("A document supplied to the model.")).describe("Documents that the model can cite while generating the response.").optional(),
-  citation_options: external_exports.looseObject({}).describe("A JSON object accepted or returned by the Cohere API.").optional(),
-  response_format: external_exports.looseObject({}).describe("A JSON object accepted or returned by the Cohere API.").optional(),
-  safety_mode: external_exports.string().describe("The Cohere safety mode to use for the request.").optional(),
-  max_tokens: external_exports.int().describe("The maximum number of output tokens the model will generate.").optional(),
-  stop_sequences: external_exports.array(external_exports.string().describe("A stop sequence.")).max(5).describe("Stop sequences that halt generation.").optional(),
-  temperature: external_exports.number().describe("Randomness used for generation.").optional(),
-  seed: external_exports.int().describe("A best-effort deterministic sampling seed.").optional(),
-  frequency_penalty: external_exports.number().describe("Penalty used to reduce repeated tokens.").optional(),
-  presence_penalty: external_exports.number().describe("Penalty used to reduce reused token content.").optional(),
-  k: external_exports.int().describe("Top-k sampling value. Use 0 to disable k-sampling.").optional(),
-  p: external_exports.number().describe("Top-p sampling value.").optional(),
-  logprobs: external_exports.boolean().describe("Whether to include generated-token log probabilities.").optional(),
-  tool_choice: external_exports.union([external_exports.string().describe("A predefined tool choice value."), external_exports.looseObject({}).describe("A JSON object accepted or returned by the Cohere API.")]).describe("Tool selection behavior for the model.").optional(),
-  thinking: external_exports.looseObject({}).describe("A JSON object accepted or returned by the Cohere API.").optional(),
-  priority: external_exports.int().describe("Lower values are handled earlier by Cohere.").optional()
-}).describe("Input for generating a synchronous response with the Cohere Chat API.");
-var chatOutput = external_exports.looseObject({
-  id: external_exports.string().describe("The unique Cohere response identifier."),
-  finish_reason: external_exports.string().describe("Why generation finished."),
-  message: external_exports.looseObject({
-    role: external_exports.string().describe("The role of the returned message.").optional(),
-    content: external_exports.array(external_exports.looseObject({}).describe("A JSON object accepted or returned by the Cohere API.")).describe("Content blocks returned by the assistant message.").optional(),
-    tool_calls: external_exports.array(external_exports.looseObject({}).describe("A JSON object accepted or returned by the Cohere API.")).describe("Tool calls returned by Cohere.").optional()
-  }).describe("The assistant message returned by Cohere."),
-  usage: external_exports.looseObject({}).describe("Usage statistics returned by Cohere.")
-}).describe("The response returned by the Cohere Chat API.");
-var embedTextsInput = external_exports.strictObject({
-  model: external_exports.string().min(1).describe("The Cohere embedding model identifier."),
-  input_type: external_exports.enum(["search_document", "search_query", "classification", "clustering"]).describe("The type of input passed to the embedding model."),
-  texts: external_exports.array(external_exports.string().describe("A text value to embed.")).min(1).max(96).describe("Text strings for the model to embed."),
-  max_tokens: external_exports.int().describe("The maximum number of tokens to embed per input.").optional(),
-  output_dimension: external_exports.int().describe("The number of dimensions for embed-v4 and newer models.").optional(),
-  embedding_types: external_exports.array(external_exports.enum(["float", "int8", "uint8", "binary", "ubinary", "base64"]).describe("The embedding representation to return.")).min(1).describe("Embedding representations to return.").optional(),
-  truncate: external_exports.enum(["NONE", "START", "END"]).describe("How Cohere handles inputs longer than the model limit.").optional(),
-  priority: external_exports.int().describe("Lower values are handled earlier by Cohere.").optional()
-}).describe("Input for generating text embeddings with the Cohere Embed API.");
-var embedTextsOutput = external_exports.looseObject({
-  id: external_exports.string().describe("The unique Cohere response identifier."),
-  embeddings: external_exports.looseObject({}).describe("Embeddings grouped by requested representation type."),
-  texts: external_exports.array(external_exports.string().describe("A text input.")).describe("Text inputs echoed by Cohere."),
-  meta: external_exports.looseObject({
-    api_version: external_exports.looseObject({
-      version: external_exports.string().describe("The API version used for the response.").optional(),
-      is_deprecated: external_exports.boolean().describe("Whether this API version is deprecated.").optional(),
-      is_experimental: external_exports.boolean().describe("Whether this API version is experimental.").optional()
-    }).describe("The Cohere API version metadata.").optional(),
-    billed_units: external_exports.looseObject({}).describe("Billing unit counts reported by Cohere.").optional(),
-    tokens: external_exports.looseObject({}).describe("Token usage counts reported by Cohere.").optional(),
-    warnings: external_exports.array(external_exports.string().describe("A warning message.")).describe("Warnings returned by Cohere.").optional()
-  }).describe("Cohere API metadata including usage and billing details.")
-}).describe("The response returned by the Cohere Embed API.");
-var rerankDocumentsInput = external_exports.strictObject({
-  model: external_exports.string().min(1).describe("The Cohere rerank model identifier."),
-  query: external_exports.string().min(1).describe("The search query used to rank documents."),
-  documents: external_exports.array(external_exports.string().describe("A text document to rank.")).min(1).describe("Texts that will be compared to the query."),
-  top_n: external_exports.int().describe("Maximum number of rerank results to return.").optional(),
-  max_tokens_per_doc: external_exports.int().describe("Maximum tokens to keep per document before ranking.").optional(),
-  priority: external_exports.int().describe("Lower values are handled earlier by Cohere.").optional()
-}).describe("Input for ranking documents by relevance with the Cohere Rerank API.");
-var rerankDocumentsOutput = external_exports.looseObject({
-  id: external_exports.string().describe("The unique Cohere response identifier."),
-  results: external_exports.array(external_exports.strictObject({
-    index: external_exports.int().describe("The original zero-based index of the ranked document.").optional(),
-    relevance_score: external_exports.number().describe("Normalized relevance score between 0 and 1.").optional()
-  }).describe("A ranked document result returned by Cohere.")).describe("Ranked document results."),
-  meta: external_exports.looseObject({
-    api_version: external_exports.looseObject({
-      version: external_exports.string().describe("The API version used for the response.").optional(),
-      is_deprecated: external_exports.boolean().describe("Whether this API version is deprecated.").optional(),
-      is_experimental: external_exports.boolean().describe("Whether this API version is experimental.").optional()
-    }).describe("The Cohere API version metadata.").optional(),
-    billed_units: external_exports.looseObject({}).describe("Billing unit counts reported by Cohere.").optional(),
-    tokens: external_exports.looseObject({}).describe("Token usage counts reported by Cohere.").optional(),
-    warnings: external_exports.array(external_exports.string().describe("A warning message.")).describe("Warnings returned by Cohere.").optional()
-  }).describe("Cohere API metadata including usage and billing details.")
-}).describe("The response returned by the Cohere Rerank API.");
-var cohereActions = {
-  chat: {
-    description: "Generate a synchronous text response using the Cohere Chat API.",
-    effect: "write",
-    inputSchema: chatInput,
-    outputSchema: external_exports.toJSONSchema(chatOutput, { io: "output", unrepresentable: "any" })
-  },
-  embed_texts: {
-    description: "Generate embeddings for text inputs using the Cohere Embed API.",
-    effect: "write",
-    inputSchema: embedTextsInput,
-    outputSchema: external_exports.toJSONSchema(embedTextsOutput, { io: "output", unrepresentable: "any" })
-  },
-  rerank_documents: {
-    description: "Rank text documents by relevance to a query using the Cohere Rerank API.",
-    effect: "write",
-    inputSchema: rerankDocumentsInput,
-    outputSchema: external_exports.toJSONSchema(rerankDocumentsOutput, { io: "output", unrepresentable: "any" })
-  }
-};
-
-// ../plugins/src/cohere/index.ts
-function createCoherePlugin() {
-  return createProviderPlugin({
-    description: "Cohere",
-    actions: cohereActions,
-    handlers: {
-      chat,
-      embed_texts: embedTexts,
-      rerank_documents: rerankDocuments
-    }
-  });
-}
-var cohere_default = createCoherePlugin();
-export {
-  createCoherePlugin,
-  cohere_default as default
-};

package/dist/deepseek-PC5YGNC5.js

@@ -1,209 +0,0 @@
-import {
-  createAuthedClient
-} from "./chunk-XYBU3WZH.js";
-import "./chunk-E6JCVKW6.js";
-import {
-  asJsonObject,
-  compactDefined,
-  trimmedText
-} from "./chunk-JXHMDBAY.js";
-import {
-  createProviderPlugin,
-  upstreamError
-} from "./chunk-ATK45BXV.js";
-import "./chunk-U7B6DS2O.js";
-import {
-  TBError,
-  external_exports
-} from "./chunk-UZM2RXZH.js";
-import "./chunk-U67V476Y.js";
-
-// ../plugins/src/deepseek/api.ts
-var SERVICE = "deepseek";
-var API_BASE = "https://api.deepseek.com";
-var ANTHROPIC_API_PREFIX = "/anthropic";
-function errorMessage(payload, bodyKind, status) {
-  const fallback = `deepseek request failed with ${status}`;
-  const body = asJsonObject(payload);
-  const nested = asJsonObject(body?.error);
-  return trimmedText(nested?.message) ?? trimmedText(body?.message) ?? (bodyKind === "invalid-json" || bodyKind === "text" ? trimmedText(payload) : void 0) ?? fallback;
-}
-function mapDeepseekError({ bodyKind, data: payload, status }) {
-  return upstreamError(status, errorMessage(payload, bodyKind, status));
-}
-var openaiHttp = createAuthedClient({
-  baseUrl: API_BASE,
-  service: SERVICE,
-  auth: { kind: "bearer" },
-  headers: { "content-type": "application/json" },
-  mapError: mapDeepseekError
-});
-var anthropicHttp = createAuthedClient({
-  baseUrl: `${API_BASE}${ANTHROPIC_API_PREFIX}/`,
-  service: SERVICE,
-  auth: { kind: "header", name: "x-api-key" },
-  headers: { "content-type": "application/json" },
-  mapError: mapDeepseekError
-});
-function assertStreamingDisabled(input) {
-  if (input.stream === true) {
-    throw new TBError("invalid_argument", "stream=true is not supported by connector actions");
-  }
-}
-async function request(ctx, options) {
-  const client = options.anthropic === true ? anthropicHttp : openaiHttp;
-  const { data } = await client.request(ctx, {
-    path: options.path,
-    method: options.method ?? "GET",
-    ...options.body === void 0 ? {} : { json: options.body },
-    invalidJsonMessage: "deepseek returned malformed JSON"
-  });
-  if (data === void 0) throw upstreamError(502, "deepseek returned malformed JSON");
-  return data;
-}
-function listModels(_input, ctx) {
-  return request(ctx, { path: "/models" });
-}
-function getUserBalance(_input, ctx) {
-  return request(ctx, { path: "/user/balance" });
-}
-function createChatCompletion(input, ctx) {
-  assertStreamingDisabled(input);
-  return request(ctx, { path: "/chat/completions", method: "POST", body: compactDefined(input) });
-}
-function createAnthropicMessage(input, ctx) {
-  assertStreamingDisabled(input);
-  return request(ctx, { path: "/v1/messages", method: "POST", body: compactDefined(input), anthropic: true });
-}
-
-// ../plugins/src/deepseek/schema.ts
-var listModelsInput = external_exports.strictObject({}).describe("No input parameters are required for this action.");
-var listModelsOutput = external_exports.strictObject({
-  object: external_exports.string().describe("The top-level object type returned by the API.").optional(),
-  data: external_exports.array(external_exports.looseObject({
-    id: external_exports.string().describe("The model identifier.").optional(),
-    object: external_exports.string().describe("The object type returned by the API.").optional(),
-    owned_by: external_exports.string().describe("The owner of the model.").optional()
-  }).describe("A DeepSeek model summary.")).describe("The list of available models.").optional()
-});
-var getUserBalanceInput = external_exports.strictObject({}).describe("No input parameters are required for this action.");
-var getUserBalanceOutput = external_exports.strictObject({
-  is_available: external_exports.boolean().describe("Whether the account balance information is currently available.").optional(),
-  balance_infos: external_exports.array(external_exports.looseObject({
-    currency: external_exports.string().describe("The currency code for the balance.").optional(),
-    total_balance: external_exports.string().describe("The total available balance.").optional(),
-    granted_balance: external_exports.string().describe("The promotional or granted balance.").optional(),
-    topped_up_balance: external_exports.string().describe("The manually topped-up balance.").optional()
-  }).describe("A balance entry returned by the DeepSeek balance API.")).describe("The list of balances grouped by currency.").optional()
-});
-var createChatCompletionInput = external_exports.strictObject({
-  model: external_exports.enum(["deepseek-chat", "deepseek-reasoner"]).describe("The DeepSeek model identifier."),
-  messages: external_exports.array(external_exports.looseObject({
-    role: external_exports.enum(["system", "user", "assistant", "tool"]).describe("The role of the message author.").optional(),
-    content: external_exports.string().describe("The text content of the message.").nullable().optional(),
-    name: external_exports.string().describe("The optional participant name for the message.").optional(),
-    prefix: external_exports.boolean().describe("Whether the assistant message should be treated as a prefix.").optional(),
-    tool_call_id: external_exports.string().describe("The identifier of the tool call that this tool message responds to.").optional(),
-    reasoning_content: external_exports.string().describe("Reasoning content provided for assistant context.").optional()
-  }).describe("A message in the OpenAI-compatible chat completion request.")).describe("The ordered conversation history to send to the model."),
-  frequency_penalty: external_exports.number().min(-2).max(2).describe("The frequency penalty to apply to repeated tokens.").optional(),
-  logprobs: external_exports.boolean().describe("Whether to include token-level log probability details in the response.").optional(),
-  max_tokens: external_exports.int().min(1).describe("The maximum number of tokens to generate.").optional(),
-  presence_penalty: external_exports.number().min(-2).max(2).describe("The presence penalty to apply to newly introduced tokens.").optional(),
-  response_format: external_exports.strictObject({
-    type: external_exports.enum(["text", "json_object"]).describe("The response format type requested from the model.").optional()
-  }).optional(),
-  stop: external_exports.union([external_exports.string().describe("A single stop sequence."), external_exports.array(external_exports.string().describe("A stop sequence.")).describe("A list of stop sequences.")]).describe("One or more sequences that stop generation.").optional(),
-  stream: external_exports.boolean().describe("Whether to request a streaming response. This connector only accepts false or omission.").optional(),
-  stream_options: external_exports.strictObject({
-    include_usage: external_exports.boolean().describe("Whether usage information should be included in stream chunks.").optional()
-  }).describe("Streaming options for the request.").optional(),
-  temperature: external_exports.number().min(0).max(2).describe("The sampling temperature to use for generation.").optional(),
-  thinking: external_exports.strictObject({
-    type: external_exports.enum(["enabled", "disabled"]).describe("Whether reasoning mode should be enabled for the request.").optional()
-  }).optional(),
-  tool_choice: external_exports.union([external_exports.enum(["none", "auto", "required"]).describe("A predefined tool selection strategy for the request."), external_exports.looseObject({
-    type: external_exports.literal("function").describe("The tool choice type. Must be function.").optional(),
-    function: external_exports.strictObject({
-      name: external_exports.string().describe("The name of the function tool to force.").optional()
-    }).optional()
-  }).describe("A tool choice that forces one specific function tool.")]).describe("The tool selection policy for the chat completion request.").optional(),
-  tools: external_exports.array(external_exports.looseObject({
-    type: external_exports.literal("function").describe("The tool type. Must be function.").optional(),
-    function: external_exports.looseObject({
-      name: external_exports.string().describe("The function name exposed to the model.").optional(),
-      description: external_exports.string().describe("A human-readable description of the function.").optional(),
-      parameters: external_exports.record(external_exports.string(), external_exports.unknown().describe("Any JSON value accepted by the DeepSeek API.")).describe("A JSON object passed through to the DeepSeek API.").optional(),
-      strict: external_exports.boolean().describe("Whether the model must follow the declared parameter schema exactly.").optional()
-    }).describe("The function definition for an OpenAI-compatible tool.").optional()
-  }).describe("An OpenAI-compatible tool definition.")).describe("The tools available to the model.").optional(),
-  top_logprobs: external_exports.int().min(0).max(20).describe("The number of top token log probabilities to return.").optional(),
-  top_p: external_exports.number().min(0).max(1).describe("The nucleus sampling threshold.").optional()
-}).describe("The input payload for the OpenAI-compatible chat completion action.");
-var createChatCompletionOutput = external_exports.looseObject({}).describe("The response payload for the chat completion action.");
-var createAnthropicMessageInput = external_exports.strictObject({
-  model: external_exports.enum(["deepseek-chat", "deepseek-reasoner"]).describe("The DeepSeek model identifier."),
-  max_tokens: external_exports.int().min(1).describe("The maximum number of tokens to generate."),
-  messages: external_exports.array(external_exports.strictObject({
-    role: external_exports.enum(["user", "assistant"]).describe("The role of the message author.").optional(),
-    content: external_exports.union([external_exports.string().describe("Plain text content."), external_exports.array(external_exports.record(external_exports.string(), external_exports.unknown().describe("Any JSON value accepted by the DeepSeek API.")).describe("A JSON object passed through to the DeepSeek API.")).describe("Structured content blocks.")]).describe("Message content as plain text or structured blocks.").optional()
-  })).describe("The ordered conversation history to send to the model."),
-  stop_sequences: external_exports.array(external_exports.string().describe("A stop sequence.")).describe("Sequences that stop generation.").optional(),
-  stream: external_exports.boolean().describe("Whether to request a streaming response. This connector only accepts false or omission.").optional(),
-  system: external_exports.union([external_exports.string().describe("Plain text content."), external_exports.array(external_exports.record(external_exports.string(), external_exports.unknown().describe("Any JSON value accepted by the DeepSeek API.")).describe("A JSON object passed through to the DeepSeek API.")).describe("Structured content blocks.")]).describe("Message content as plain text or structured blocks.").optional(),
-  temperature: external_exports.number().min(0).max(2).describe("The sampling temperature to use for generation.").optional(),
-  thinking: external_exports.looseObject({
-    type: external_exports.string().describe("The thinking mode requested by the Anthropic-compatible API.").optional(),
-    budget_tokens: external_exports.int().describe("The maximum number of tokens allocated to thinking.").optional()
-  }).describe("Thinking configuration for the Anthropic-compatible request.").optional(),
-  tool_choice: external_exports.union([external_exports.enum(["auto", "any", "none"]).describe("A predefined Anthropic-compatible tool choice mode."), external_exports.record(external_exports.string(), external_exports.unknown().describe("Any JSON value accepted by the DeepSeek API.")).describe("A JSON object passed through to the DeepSeek API.")]).describe("How the model should choose tools for the request.").optional(),
-  tools: external_exports.array(external_exports.record(external_exports.string(), external_exports.unknown().describe("Any JSON value accepted by the DeepSeek API.")).describe("A JSON object passed through to the DeepSeek API.")).describe("The tools available to the model.").optional(),
-  top_p: external_exports.number().describe("The nucleus sampling threshold.").optional()
-}).describe("The input payload for the Anthropic-compatible message action.");
-var createAnthropicMessageOutput = external_exports.looseObject({}).describe("The response payload for the Anthropic-compatible message action.");
-var deepseekActions = {
-  list_models: {
-    description: "List the available DeepSeek models.",
-    effect: "read",
-    inputSchema: listModelsInput,
-    outputSchema: external_exports.toJSONSchema(listModelsOutput, { io: "output", unrepresentable: "any" })
-  },
-  get_user_balance: {
-    description: "Get the current DeepSeek account balance.",
-    effect: "read",
-    inputSchema: getUserBalanceInput,
-    outputSchema: external_exports.toJSONSchema(getUserBalanceOutput, { io: "output", unrepresentable: "any" })
-  },
-  create_chat_completion: {
-    description: "Create a DeepSeek chat completion via the OpenAI-compatible API.",
-    effect: "write",
-    inputSchema: createChatCompletionInput,
-    outputSchema: external_exports.toJSONSchema(createChatCompletionOutput, { io: "output", unrepresentable: "any" })
-  },
-  create_anthropic_message: {
-    description: "Create a DeepSeek message via the Anthropic-compatible API.",
-    effect: "write",
-    inputSchema: createAnthropicMessageInput,
-    outputSchema: external_exports.toJSONSchema(createAnthropicMessageOutput, { io: "output", unrepresentable: "any" })
-  }
-};
-
-// ../plugins/src/deepseek/index.ts
-function createDeepseekPlugin() {
-  return createProviderPlugin({
-    description: "DeepSeek",
-    credentialProbe: "list_models",
-    actions: deepseekActions,
-    handlers: {
-      list_models: listModels,
-      get_user_balance: getUserBalance,
-      create_chat_completion: createChatCompletion,
-      create_anthropic_message: createAnthropicMessage
-    }
-  });
-}
-var deepseek_default = createDeepseekPlugin();
-export {
-  createDeepseekPlugin,
-  deepseek_default as default
-};