@codeany/open-agent-sdk 0.2.2 → 0.2.3

package/src/types.ts

@@ -1,486 +0,0 @@
-/**
- * Core type definitions for the Agent SDK
- */
-
-// Content block types (provider-agnostic, compatible with Anthropic format)
-export type ContentBlockParam =
-  | { type: 'text'; text: string }
-  | { type: 'image'; source: any }
-  | { type: 'tool_use'; id: string; name: string; input: any }
-  | { type: 'tool_result'; tool_use_id: string; content: string | any[]; is_error?: boolean }
-
-export type ContentBlock =
-  | { type: 'text'; text: string }
-  | { type: 'tool_use'; id: string; name: string; input: any }
-  | { type: 'thinking'; thinking: string }
-
-// --------------------------------------------------------------------------
-// Message Types
-// --------------------------------------------------------------------------
-
-export type MessageRole = 'user' | 'assistant'
-
-export interface ConversationMessage {
-  role: MessageRole
-  content: string | ContentBlockParam[]
-}
-
-export interface UserMessage {
-  type: 'user'
-  message: ConversationMessage
-  uuid: string
-  timestamp: string
-}
-
-export interface AssistantMessage {
-  type: 'assistant'
-  message: {
-    role: 'assistant'
-    content: ContentBlock[]
-  }
-  uuid: string
-  timestamp: string
-  usage?: TokenUsage
-  cost?: number
-}
-
-export type Message = UserMessage | AssistantMessage
-
-// --------------------------------------------------------------------------
-// SDK Message Types (streaming events)
-// --------------------------------------------------------------------------
-
-export type SDKMessage =
-  | SDKAssistantMessage
-  | SDKToolResultMessage
-  | SDKResultMessage
-  | SDKPartialMessage
-  | SDKSystemMessage
-  | SDKCompactBoundaryMessage
-  | SDKStatusMessage
-  | SDKTaskNotificationMessage
-  | SDKRateLimitEvent
-
-export interface SDKAssistantMessage {
-  type: 'assistant'
-  uuid?: string
-  session_id?: string
-  message: {
-    role: 'assistant'
-    content: ContentBlock[]
-  }
-  parent_tool_use_id?: string | null
-}
-
-export interface SDKToolResultMessage {
-  type: 'tool_result'
-  result: {
-    tool_use_id: string
-    tool_name: string
-    output: string
-  }
-}
-
-export interface SDKResultMessage {
-  type: 'result'
-  subtype: 'success' | 'error_max_turns' | 'error_during_execution' | 'error_max_budget_usd' | string
-  uuid?: string
-  session_id?: string
-  is_error?: boolean
-  num_turns?: number
-  result?: string
-  stop_reason?: string | null
-  total_cost_usd?: number
-  duration_ms?: number
-  duration_api_ms?: number
-  usage?: TokenUsage
-  model_usage?: Record<string, { input_tokens: number; output_tokens: number }>
-  permission_denials?: Array<{ tool: string; reason: string }>
-  structured_output?: unknown
-  errors?: string[]
-  /** @deprecated Use total_cost_usd */
-  cost?: number
-}
-
-export interface SDKPartialMessage {
-  type: 'partial_message'
-  partial: {
-    type: 'text' | 'tool_use'
-    text?: string
-    name?: string
-    input?: string
-  }
-}
-
-/** Emitted once at session start with initialization info. */
-export interface SDKSystemMessage {
-  type: 'system'
-  subtype: 'init'
-  uuid?: string
-  session_id: string
-  tools: string[]
-  model: string
-  cwd: string
-  mcp_servers: Array<{ name: string; status: string }>
-  permission_mode: string
-}
-
-/** Marks a compaction boundary in the conversation. */
-export interface SDKCompactBoundaryMessage {
-  type: 'system'
-  subtype: 'compact_boundary'
-  summary?: string
-}
-
-/** Status update during long operations. */
-export interface SDKStatusMessage {
-  type: 'system'
-  subtype: 'status'
-  message: string
-}
-
-/** Task lifecycle notification. */
-export interface SDKTaskNotificationMessage {
-  type: 'system'
-  subtype: 'task_notification'
-  task_id: string
-  status: string
-  message?: string
-}
-
-/** Rate limit event. */
-export interface SDKRateLimitEvent {
-  type: 'system'
-  subtype: 'rate_limit'
-  retry_after_ms?: number
-  message: string
-}
-
-// --------------------------------------------------------------------------
-// Token Usage
-// --------------------------------------------------------------------------
-
-export interface TokenUsage {
-  input_tokens: number
-  output_tokens: number
-  cache_creation_input_tokens?: number
-  cache_read_input_tokens?: number
-}
-
-// --------------------------------------------------------------------------
-// Tool Types
-// --------------------------------------------------------------------------
-
-export interface ToolDefinition {
-  name: string
-  description: string
-  inputSchema: ToolInputSchema
-  call: (input: any, context: ToolContext) => Promise<ToolResult>
-  isReadOnly?: () => boolean
-  isConcurrencySafe?: () => boolean
-  isEnabled?: () => boolean
-  prompt?: (context: ToolContext) => Promise<string>
-}
-
-export interface ToolInputSchema {
-  type: 'object'
-  properties: Record<string, any>
-  required?: string[]
-}
-
-export interface ToolContext {
-  cwd: string
-  abortSignal?: AbortSignal
-  /** Parent agent's LLM provider (inherited by subagents) */
-  provider?: import('./providers/types.js').LLMProvider
-  /** Parent agent's model ID */
-  model?: string
-  /** Parent agent's API type */
-  apiType?: import('./providers/types.js').ApiType
-}
-
-export interface ToolResult {
-  type: 'tool_result'
-  tool_use_id: string
-  content: string | any[]
-  is_error?: boolean
-}
-
-// --------------------------------------------------------------------------
-// Permission Types
-// --------------------------------------------------------------------------
-
-export type PermissionMode =
-  | 'default'
-  | 'acceptEdits'
-  | 'bypassPermissions'
-  | 'plan'
-  | 'dontAsk'
-  | 'auto'
-
-export type PermissionBehavior = 'allow' | 'deny'
-
-export type CanUseToolResult = {
-  behavior: PermissionBehavior
-  updatedInput?: unknown
-  message?: string
-}
-
-export type CanUseToolFn = (
-  tool: ToolDefinition,
-  input: unknown,
-) => Promise<CanUseToolResult>
-
-// --------------------------------------------------------------------------
-// MCP Types
-// --------------------------------------------------------------------------
-
-export type McpServerConfig =
-  | McpStdioConfig
-  | McpSseConfig
-  | McpHttpConfig
-
-export interface McpStdioConfig {
-  type?: 'stdio'
-  command: string
-  args?: string[]
-  env?: Record<string, string>
-}
-
-export interface McpSseConfig {
-  type: 'sse'
-  url: string
-  headers?: Record<string, string>
-}
-
-export interface McpHttpConfig {
-  type: 'http'
-  url: string
-  headers?: Record<string, string>
-}
-
-// --------------------------------------------------------------------------
-// Agent Types
-// --------------------------------------------------------------------------
-
-export interface AgentDefinition {
-  description: string
-  prompt: string
-  tools?: string[]
-  disallowedTools?: string[]
-  model?: 'sonnet' | 'opus' | 'haiku' | 'inherit' | string
-  mcpServers?: Array<string | { name: string; tools?: string[] }>
-  skills?: string[]
-  maxTurns?: number
-  criticalSystemReminder_EXPERIMENTAL?: string
-}
-
-export interface ThinkingConfig {
-  type: 'adaptive' | 'enabled' | 'disabled'
-  budgetTokens?: number
-}
-
-// --------------------------------------------------------------------------
-// Sandbox Types
-// --------------------------------------------------------------------------
-
-export interface SandboxSettings {
-  enabled?: boolean
-  autoAllowBashIfSandboxed?: boolean
-  excludedCommands?: string[]
-  allowUnsandboxedCommands?: boolean
-  network?: SandboxNetworkConfig
-  filesystem?: SandboxFilesystemConfig
-  ignoreViolations?: Record<string, string[]>
-  enableWeakerNestedSandbox?: boolean
-  ripgrep?: { command: string; args?: string[] }
-}
-
-export interface SandboxNetworkConfig {
-  allowedDomains?: string[]
-  allowManagedDomainsOnly?: boolean
-  allowLocalBinding?: boolean
-  allowUnixSockets?: string[]
-  allowAllUnixSockets?: boolean
-  httpProxyPort?: number
-  socksProxyPort?: number
-}
-
-export interface SandboxFilesystemConfig {
-  allowWrite?: string[]
-  denyWrite?: string[]
-  denyRead?: string[]
-}
-
-// --------------------------------------------------------------------------
-// Output Format
-// --------------------------------------------------------------------------
-
-export interface OutputFormat {
-  type: 'json_schema'
-  schema: Record<string, unknown>
-}
-
-// --------------------------------------------------------------------------
-// Setting Sources
-// --------------------------------------------------------------------------
-
-export type SettingSource = 'user' | 'project' | 'local'
-
-// --------------------------------------------------------------------------
-// Model Info
-// --------------------------------------------------------------------------
-
-export interface ModelInfo {
-  value: string
-  displayName: string
-  description: string
-  supportsEffort?: boolean
-  supportedEffortLevels?: ('low' | 'medium' | 'high' | 'max')[]
-  supportsAdaptiveThinking?: boolean
-  supportsFastMode?: boolean
-}
-
-export interface AgentOptions {
-  /** LLM model ID */
-  model?: string
-  /**
-   * API type: 'anthropic-messages' or 'openai-completions'.
-   * Falls back to CODEANY_API_TYPE env var. Default: 'anthropic-messages'.
-   */
-  apiType?: import('./providers/types.js').ApiType
-  /** API key. Falls back to CODEANY_API_KEY env var. */
-  apiKey?: string
-  /** API base URL override */
-  baseURL?: string
-  /** Working directory for file/shell tools */
-  cwd?: string
-  /** System prompt override or preset */
-  systemPrompt?: string | { type: 'preset'; preset: 'default'; append?: string }
-  /** Append to default system prompt */
-  appendSystemPrompt?: string
-  /** Available tools (ToolDefinition[] or string[] preset) */
-  tools?: ToolDefinition[] | string[] | { type: 'preset'; preset: 'default' }
-  /** Maximum number of agentic turns per query */
-  maxTurns?: number
-  /** Maximum USD budget per query */
-  maxBudgetUsd?: number
-  /** Extended thinking configuration */
-  thinking?: ThinkingConfig
-  /** Maximum thinking tokens (deprecated, use thinking.budgetTokens) */
-  maxThinkingTokens?: number
-  /** Structured output JSON schema */
-  jsonSchema?: Record<string, unknown>
-  /** Structured output format */
-  outputFormat?: OutputFormat
-  /** Permission handler callback */
-  canUseTool?: CanUseToolFn
-  /** Permission mode controlling tool approval behavior */
-  permissionMode?: PermissionMode
-  /** Abort controller for cancellation */
-  abortController?: AbortController
-  /** Abort signal for cancellation */
-  abortSignal?: AbortSignal
-  /** Whether to include partial streaming events */
-  includePartialMessages?: boolean
-  /** Environment variables */
-  env?: Record<string, string | undefined>
-  /** Tool names to pre-approve without prompting */
-  allowedTools?: string[]
-  /** Tool names to deny */
-  disallowedTools?: string[]
-  /** MCP server configurations */
-  mcpServers?: Record<string, McpServerConfig | any> // supports McpSdkServerConfig
-  /** Custom subagent definitions */
-  agents?: Record<string, AgentDefinition>
-  /** Maximum tokens for responses */
-  maxTokens?: number
-  /** Effort level for reasoning */
-  effort?: 'low' | 'medium' | 'high' | 'max'
-  /** Fallback model if primary is unavailable */
-  fallbackModel?: string
-  /** Continue the most recent session in cwd */
-  continue?: boolean
-  /** Resume a specific session by ID */
-  resume?: string
-  /** Fork a session instead of continuing it */
-  forkSession?: boolean
-  /** Persist session to disk */
-  persistSession?: boolean
-  /** Explicit session ID */
-  sessionId?: string
-  /** Enable file checkpointing (for rewindFiles) */
-  enableFileCheckpointing?: boolean
-  /** Sandbox configuration */
-  sandbox?: SandboxSettings
-  /** Load settings from filesystem */
-  settingSources?: SettingSource[]
-  /** Plugin configurations */
-  plugins?: Array<{ name: string; config?: Record<string, unknown> }>
-  /** Additional working directories */
-  additionalDirectories?: string[]
-  /** Default agent to use */
-  agent?: string
-  /** Debug mode */
-  debug?: boolean
-  /** Debug log file */
-  debugFile?: string
-  /** Tool-specific configuration */
-  toolConfig?: Record<string, unknown>
-  /** Enable prompt suggestions */
-  promptSuggestions?: boolean
-  /** Strict MCP config validation */
-  strictMcpConfig?: boolean
-  /** Extra CLI arguments */
-  extraArgs?: Record<string, string | null>
-  /** SDK betas to enable */
-  betas?: string[]
-  /** Permission prompt tool name override */
-  permissionPromptToolName?: string
-  /** Hook configurations */
-  hooks?: Record<string, Array<{
-    matcher?: string
-    hooks: Array<(input: any, toolUseId: string, context: { signal: AbortSignal }) => Promise<any>>
-    timeout?: number
-  }>>
-}
-
-export interface QueryResult {
-  /** Final text output from the assistant */
-  text: string
-  /** Token usage */
-  usage: TokenUsage
-  /** Number of agentic turns */
-  num_turns: number
-  /** Duration in milliseconds */
-  duration_ms: number
-  /** All conversation messages */
-  messages: Message[]
-}
-
-// --------------------------------------------------------------------------
-// Query Engine Types
-// --------------------------------------------------------------------------
-
-export interface QueryEngineConfig {
-  cwd: string
-  model: string
-  /** LLM provider instance (created from apiType) */
-  provider: import('./providers/types.js').LLMProvider
-  tools: ToolDefinition[]
-  systemPrompt?: string
-  appendSystemPrompt?: string
-  maxTurns: number
-  maxBudgetUsd?: number
-  maxTokens: number
-  thinking?: ThinkingConfig
-  jsonSchema?: Record<string, unknown>
-  canUseTool: CanUseToolFn
-  includePartialMessages: boolean
-  abortSignal?: AbortSignal
-  agents?: Record<string, AgentDefinition>
-  /** Hook registry for lifecycle events */
-  hookRegistry?: import('./hooks.js').HookRegistry
-  /** Session ID for hook context */
-  sessionId?: string
-}

package/src/utils/compact.ts

@@ -1,207 +0,0 @@
-/**
- * Context Compression / Auto-Compaction
- *
- * Summarizes long conversation histories when context window fills up.
- * Three-tier system:
- * 1. Auto-compact: triggered when tokens exceed threshold
- * 2. Micro-compact: cache-aware per-request optimization
- * 3. Session memory compaction: consolidates across sessions
- */
-
-import type { LLMProvider } from '../providers/types.js'
-import type { NormalizedMessageParam } from '../providers/types.js'
-import {
-  estimateMessagesTokens,
-  getAutoCompactThreshold,
-} from './tokens.js'
-
-/**
- * State for tracking auto-compaction across turns.
- */
-export interface AutoCompactState {
-  compacted: boolean
-  turnCounter: number
-  consecutiveFailures: number
-}
-
-/**
- * Create initial auto-compact state.
- */
-export function createAutoCompactState(): AutoCompactState {
-  return {
-    compacted: false,
-    turnCounter: 0,
-    consecutiveFailures: 0,
-  }
-}
-
-/**
- * Check if auto-compaction should trigger.
- */
-export function shouldAutoCompact(
-  messages: any[],
-  model: string,
-  state: AutoCompactState,
-): boolean {
-  if (state.consecutiveFailures >= 3) return false
-
-  const estimatedTokens = estimateMessagesTokens(messages)
-  const threshold = getAutoCompactThreshold(model)
-
-  return estimatedTokens >= threshold
-}
-
-/**
- * Compact conversation by summarizing with the LLM.
- *
- * Sends the entire conversation to the LLM for summarization,
- * then replaces the history with a compact summary.
- */
-export async function compactConversation(
-  provider: LLMProvider,
-  model: string,
-  messages: any[],
-  state: AutoCompactState,
-): Promise<{
-  compactedMessages: NormalizedMessageParam[]
-  summary: string
-  state: AutoCompactState
-}> {
-  try {
-    // Strip images before compacting to save tokens
-    const strippedMessages = stripImagesFromMessages(messages)
-
-    // Build compaction prompt
-    const compactionPrompt = buildCompactionPrompt(strippedMessages)
-
-    const response = await provider.createMessage({
-      model,
-      maxTokens: 8192,
-      system: 'You are a conversation summarizer. Create a detailed summary of the conversation that preserves all important context, decisions made, files modified, tool outputs, and current state. The summary should allow the conversation to continue seamlessly.',
-      messages: [
-        {
-          role: 'user',
-          content: compactionPrompt,
-        },
-      ],
-    })
-
-    const summary = response.content
-      .filter((b) => b.type === 'text')
-      .map((b) => (b as { type: 'text'; text: string }).text)
-      .join('\n')
-
-    // Replace messages with summary
-    const compactedMessages: NormalizedMessageParam[] = [
-      {
-        role: 'user',
-        content: `[Previous conversation summary]\n\n${summary}\n\n[End of summary - conversation continues below]`,
-      },
-      {
-        role: 'assistant',
-        content: 'I understand the context from the previous conversation. I\'ll continue from where we left off.',
-      },
-    ]
-
-    return {
-      compactedMessages,
-      summary,
-      state: {
-        compacted: true,
-        turnCounter: state.turnCounter,
-        consecutiveFailures: 0,
-      },
-    }
-  } catch (err: any) {
-    return {
-      compactedMessages: messages,
-      summary: '',
-      state: {
-        ...state,
-        consecutiveFailures: state.consecutiveFailures + 1,
-      },
-    }
-  }
-}
-
-/**
- * Strip images from messages for compaction safety.
- */
-function stripImagesFromMessages(
-  messages: any[],
-): any[] {
-  return messages.map((msg: any) => {
-    if (typeof msg.content === 'string') return msg
-
-    const filtered = (msg.content as any[]).filter((block: any) => {
-      return block.type !== 'image'
-    })
-
-    return { ...msg, content: filtered.length > 0 ? filtered : '[content removed for compaction]' }
-  })
-}
-
-/**
- * Build compaction prompt from messages.
- */
-function buildCompactionPrompt(messages: any[]): string {
-  const parts: string[] = ['Please summarize this conversation:\n']
-
-  for (const msg of messages) {
-    const role = msg.role === 'user' ? 'User' : 'Assistant'
-
-    if (typeof msg.content === 'string') {
-      parts.push(`${role}: ${msg.content.slice(0, 5000)}`)
-    } else if (Array.isArray(msg.content)) {
-      const texts: string[] = []
-      for (const block of msg.content as any[]) {
-        if (block.type === 'text') {
-          texts.push(block.text.slice(0, 3000))
-        } else if (block.type === 'tool_use') {
-          texts.push(`[Tool: ${block.name}]`)
-        } else if (block.type === 'tool_result') {
-          const content = typeof block.content === 'string'
-            ? block.content.slice(0, 1000)
-            : '[tool result]'
-          texts.push(`[Tool Result: ${content}]`)
-        }
-      }
-      if (texts.length > 0) {
-        parts.push(`${role}: ${texts.join('\n')}`)
-      }
-    }
-  }
-
-  return parts.join('\n\n')
-}
-
-/**
- * Micro-compact: optimize messages by truncating large tool results
- * to fit within token budgets.
- */
-export function microCompactMessages(
-  messages: any[],
-  maxToolResultChars: number = 50000,
-): any[] {
-  return messages.map((msg: any) => {
-    if (typeof msg.content === 'string') return msg
-    if (!Array.isArray(msg.content)) return msg
-
-    const content = (msg.content as any[]).map((block: any) => {
-      if (block.type === 'tool_result' && typeof block.content === 'string') {
-        if (block.content.length > maxToolResultChars) {
-          return {
-            ...block,
-            content:
-              block.content.slice(0, maxToolResultChars / 2) +
-              '\n...(truncated)...\n' +
-              block.content.slice(-maxToolResultChars / 2),
-          }
-        }
-      }
-      return block
-    })
-
-    return { ...msg, content }
-  })
-}