@d34dman/flowdrop 0.0.56 → 0.0.58

package/dist/adapters/agentspec/defaultNodeTypes.js

@@ -0,0 +1,561 @@
+/**
+ * Agent Spec Default Node Types
+ *
+ * Provides optional starter node type definitions for the Agent Spec format.
+ * These are full NodeMetadata objects with visual type, category, default ports,
+ * config schema, and icon — suitable for populating the sidebar.
+ *
+ * These definitions are NOT required by the adapter — the adapter uses
+ * componentTypeDefaults.ts for import/export infrastructure.
+ * Users can provide their own node definitions instead.
+ *
+ * @example
+ * ```typescript
+ * import { getDefaultAgentSpecNodeTypes } from '@d34dman/flowdrop/core';
+ *
+ * mountFlowDropApp(container, {
+ *   nodes: getDefaultAgentSpecNodeTypes(), // or your own definitions
+ * });
+ * ```
+ */
+import { AGENTSPEC_NAMESPACE as AGENTSPEC_NS, TRIGGER_INPUT, TRIGGER_OUTPUT } from './componentTypeDefaults.js';
+/**
+ * Build the full registry of Agent Spec node types.
+ */
+function buildRegistry() {
+    const registry = new Map();
+    // ========================================================================
+    // StartNode — Graph entry point
+    // ========================================================================
+    registry.set('start_node', {
+        componentType: 'start_node',
+        metadata: {
+            id: `${AGENTSPEC_NS}.start_node`,
+            name: 'Start',
+            type: 'terminal',
+            description: 'Flow entry point. Defines the initial inputs for the flow.',
+            category: 'triggers',
+            version: '1.0.0',
+            icon: 'mdi:play-circle',
+            color: '#22c55e',
+            badge: 'START',
+            inputs: [],
+            outputs: [TRIGGER_OUTPUT],
+            configSchema: {
+                type: 'object',
+                properties: {}
+            },
+            formats: ['agentspec'],
+            extensions: { 'agentspec:component_type': 'start_node' }
+        }
+    });
+    // ========================================================================
+    // EndNode — Graph exit point
+    // ========================================================================
+    registry.set('end_node', {
+        componentType: 'end_node',
+        metadata: {
+            id: `${AGENTSPEC_NS}.end_node`,
+            name: 'End',
+            type: 'terminal',
+            description: 'Flow exit point. Defines the final outputs of the flow.',
+            category: 'outputs',
+            version: '1.0.0',
+            icon: 'mdi:stop-circle',
+            color: '#ef4444',
+            badge: 'END',
+            inputs: [TRIGGER_INPUT],
+            outputs: [],
+            configSchema: {
+                type: 'object',
+                properties: {}
+            },
+            formats: ['agentspec'],
+            extensions: { 'agentspec:component_type': 'end_node' }
+        }
+    });
+    // ========================================================================
+    // LLMNode — Text generation from prompts
+    // ========================================================================
+    registry.set('llm_node', {
+        componentType: 'llm_node',
+        metadata: {
+            id: `${AGENTSPEC_NS}.llm_node`,
+            name: 'LLM',
+            type: 'default',
+            description: 'Generate text using a large language model with configurable prompts.',
+            category: 'ai',
+            version: '1.0.0',
+            icon: 'mdi:brain',
+            color: '#8b5cf6',
+            badge: 'LLM',
+            inputs: [
+                TRIGGER_INPUT,
+                {
+                    id: 'prompt',
+                    name: 'Prompt',
+                    type: 'input',
+                    dataType: 'string',
+                    required: false,
+                    description: 'Input prompt text (can also use {{variable}} in prompt template)'
+                }
+            ],
+            outputs: [
+                TRIGGER_OUTPUT,
+                {
+                    id: 'llm_output',
+                    name: 'LLM Output',
+                    type: 'output',
+                    dataType: 'string',
+                    description: 'Generated text from the LLM'
+                }
+            ],
+            configSchema: {
+                type: 'object',
+                properties: {
+                    prompt_template: {
+                        type: 'string',
+                        title: 'Prompt Template',
+                        description: 'Prompt template with {{variable}} placeholders for dynamic inputs',
+                        format: 'multiline',
+                        default: '{{prompt}}'
+                    },
+                    system_prompt: {
+                        type: 'string',
+                        title: 'System Prompt',
+                        description: 'System message to set the LLM behavior',
+                        format: 'multiline',
+                        default: ''
+                    },
+                    llm_config_ref: {
+                        type: 'string',
+                        title: 'LLM Configuration',
+                        description: 'Reference to an LLM configuration (e.g., model name, provider)',
+                        default: ''
+                    }
+                }
+            },
+            formats: ['agentspec'],
+            extensions: { 'agentspec:component_type': 'llm_node' }
+        }
+    });
+    // ========================================================================
+    // BranchingNode — Conditional routing
+    // ========================================================================
+    registry.set('branching_node', {
+        componentType: 'branching_node',
+        metadata: {
+            id: `${AGENTSPEC_NS}.branching_node`,
+            name: 'Branch',
+            type: 'gateway',
+            description: 'Route execution to different paths based on conditions.',
+            category: 'logic',
+            version: '1.0.0',
+            icon: 'mdi:source-branch',
+            color: '#f59e0b',
+            badge: 'BRANCH',
+            inputs: [
+                TRIGGER_INPUT,
+                {
+                    id: 'input',
+                    name: 'Input',
+                    type: 'input',
+                    dataType: 'mixed',
+                    required: false,
+                    description: 'Data to evaluate branch conditions against'
+                }
+            ],
+            outputs: [],
+            configSchema: {
+                type: 'object',
+                properties: {
+                    branches: {
+                        type: 'array',
+                        title: 'Branches',
+                        description: 'Conditional branches for routing',
+                        items: {
+                            type: 'object',
+                            properties: {
+                                name: {
+                                    type: 'string',
+                                    title: 'Branch ID',
+                                    description: 'Unique identifier for the branch'
+                                },
+                                label: {
+                                    type: 'string',
+                                    title: 'Label',
+                                    description: 'Display label for the branch'
+                                },
+                                condition: {
+                                    type: 'string',
+                                    title: 'Condition',
+                                    description: 'Expression that determines when this branch activates'
+                                },
+                                isDefault: {
+                                    type: 'boolean',
+                                    title: 'Default Branch',
+                                    description: 'Whether this is the fallback branch',
+                                    default: false
+                                }
+                            },
+                            required: ['name']
+                        }
+                    }
+                }
+            },
+            formats: ['agentspec'],
+            extensions: { 'agentspec:component_type': 'branching_node' }
+        }
+    });
+    // ========================================================================
+    // ToolNode — Tool execution
+    // ========================================================================
+    registry.set('tool_node', {
+        componentType: 'tool_node',
+        metadata: {
+            id: `${AGENTSPEC_NS}.tool_node`,
+            name: 'Tool',
+            type: 'tool',
+            description: 'Execute a tool function with inputs and receive outputs.',
+            category: 'tools',
+            version: '1.0.0',
+            icon: 'mdi:wrench',
+            color: '#06b6d4',
+            badge: 'TOOL',
+            inputs: [
+                TRIGGER_INPUT,
+                {
+                    id: 'tool',
+                    name: 'Tool',
+                    type: 'input',
+                    dataType: 'tool',
+                    required: false,
+                    description: 'Tool connection'
+                }
+            ],
+            outputs: [
+                TRIGGER_OUTPUT,
+                {
+                    id: 'tool',
+                    name: 'Tool',
+                    type: 'output',
+                    dataType: 'tool',
+                    description: 'Tool passthrough'
+                },
+                {
+                    id: 'result',
+                    name: 'Result',
+                    type: 'output',
+                    dataType: 'mixed',
+                    description: 'Tool execution result'
+                }
+            ],
+            configSchema: {
+                type: 'object',
+                properties: {
+                    tool_ref: {
+                        type: 'string',
+                        title: 'Tool Reference',
+                        description: 'Reference to the tool to execute ($component_ref:tool_name)',
+                        default: ''
+                    }
+                }
+            },
+            formats: ['agentspec'],
+            extensions: { 'agentspec:component_type': 'tool_node' }
+        }
+    });
+    // ========================================================================
+    // APINode — API call
+    // ========================================================================
+    registry.set('api_node', {
+        componentType: 'api_node',
+        metadata: {
+            id: `${AGENTSPEC_NS}.api_node`,
+            name: 'API Call',
+            type: 'default',
+            description: 'Make an HTTP API call with configurable endpoint, method, and headers.',
+            category: 'data',
+            version: '1.0.0',
+            icon: 'mdi:api',
+            color: '#3b82f6',
+            badge: 'API',
+            inputs: [
+                TRIGGER_INPUT,
+                {
+                    id: 'body',
+                    name: 'Request Body',
+                    type: 'input',
+                    dataType: 'json',
+                    required: false,
+                    description: 'Request body data'
+                }
+            ],
+            outputs: [
+                TRIGGER_OUTPUT,
+                {
+                    id: 'response',
+                    name: 'Response',
+                    type: 'output',
+                    dataType: 'json',
+                    description: 'API response data'
+                },
+                {
+                    id: 'status_code',
+                    name: 'Status Code',
+                    type: 'output',
+                    dataType: 'number',
+                    description: 'HTTP status code'
+                }
+            ],
+            configSchema: {
+                type: 'object',
+                properties: {
+                    endpoint: {
+                        type: 'string',
+                        title: 'Endpoint URL',
+                        description: 'The API endpoint to call',
+                        default: ''
+                    },
+                    method: {
+                        type: 'string',
+                        title: 'HTTP Method',
+                        enum: ['GET', 'POST', 'PUT', 'PATCH', 'DELETE'],
+                        default: 'GET'
+                    },
+                    headers: {
+                        type: 'string',
+                        title: 'Headers',
+                        description: 'Request headers as JSON',
+                        format: 'multiline',
+                        default: '{}'
+                    }
+                }
+            },
+            formats: ['agentspec'],
+            extensions: { 'agentspec:component_type': 'api_node' }
+        }
+    });
+    // ========================================================================
+    // AgentNode — Multi-round agent conversation
+    // ========================================================================
+    registry.set('agent_node', {
+        componentType: 'agent_node',
+        metadata: {
+            id: `${AGENTSPEC_NS}.agent_node`,
+            name: 'Agent',
+            type: 'default',
+            description: 'Run a multi-round agent conversation within the flow.',
+            category: 'agents',
+            version: '1.0.0',
+            icon: 'mdi:robot',
+            color: '#ec4899',
+            badge: 'AGENT',
+            inputs: [
+                TRIGGER_INPUT,
+                {
+                    id: 'input',
+                    name: 'Input',
+                    type: 'input',
+                    dataType: 'string',
+                    required: false,
+                    description: 'Input message for the agent'
+                }
+            ],
+            outputs: [
+                TRIGGER_OUTPUT,
+                {
+                    id: 'output',
+                    name: 'Output',
+                    type: 'output',
+                    dataType: 'string',
+                    description: 'Agent response output'
+                }
+            ],
+            configSchema: {
+                type: 'object',
+                properties: {
+                    agent_ref: {
+                        type: 'string',
+                        title: 'Agent Reference',
+                        description: 'Reference to the agent ($component_ref:agent_name)',
+                        default: ''
+                    },
+                    max_turns: {
+                        type: 'integer',
+                        title: 'Max Turns',
+                        description: 'Maximum number of conversation turns',
+                        minimum: 1,
+                        default: 10
+                    }
+                }
+            },
+            formats: ['agentspec'],
+            extensions: { 'agentspec:component_type': 'agent_node' }
+        }
+    });
+    // ========================================================================
+    // FlowNode — Nested flow execution
+    // ========================================================================
+    registry.set('flow_node', {
+        componentType: 'flow_node',
+        metadata: {
+            id: `${AGENTSPEC_NS}.flow_node`,
+            name: 'Sub-Flow',
+            type: 'simple',
+            description: 'Execute another flow as a sub-routine within this flow.',
+            category: 'processing',
+            version: '1.0.0',
+            icon: 'mdi:sitemap',
+            color: '#14b8a6',
+            badge: 'FLOW',
+            inputs: [
+                TRIGGER_INPUT,
+                {
+                    id: 'input',
+                    name: 'Input',
+                    type: 'input',
+                    dataType: 'mixed',
+                    required: false,
+                    description: 'Input data passed to the sub-flow'
+                }
+            ],
+            outputs: [
+                TRIGGER_OUTPUT,
+                {
+                    id: 'output',
+                    name: 'Output',
+                    type: 'output',
+                    dataType: 'mixed',
+                    description: 'Output data from the sub-flow'
+                }
+            ],
+            configSchema: {
+                type: 'object',
+                properties: {
+                    flow_ref: {
+                        type: 'string',
+                        title: 'Flow Reference',
+                        description: 'Reference to the flow ($component_ref:flow_name)',
+                        default: ''
+                    }
+                }
+            },
+            formats: ['agentspec'],
+            extensions: { 'agentspec:component_type': 'flow_node' }
+        }
+    });
+    // ========================================================================
+    // MapNode — Map-reduce operations
+    // ========================================================================
+    registry.set('map_node', {
+        componentType: 'map_node',
+        metadata: {
+            id: `${AGENTSPEC_NS}.map_node`,
+            name: 'Map',
+            type: 'default',
+            description: 'Apply a flow or operation to each item in a collection (map-reduce).',
+            category: 'processing',
+            version: '1.0.0',
+            icon: 'mdi:map-marker-path',
+            color: '#f97316',
+            badge: 'MAP',
+            inputs: [
+                TRIGGER_INPUT,
+                {
+                    id: 'collection',
+                    name: 'Collection',
+                    type: 'input',
+                    dataType: 'array',
+                    required: true,
+                    description: 'Input collection to iterate over'
+                }
+            ],
+            outputs: [
+                TRIGGER_OUTPUT,
+                {
+                    id: 'results',
+                    name: 'Results',
+                    type: 'output',
+                    dataType: 'array',
+                    description: 'Mapped output collection'
+                }
+            ],
+            configSchema: {
+                type: 'object',
+                properties: {
+                    input_collection: {
+                        type: 'string',
+                        title: 'Input Collection Property',
+                        description: 'Name of the input property containing the collection',
+                        default: 'collection'
+                    },
+                    output_collection: {
+                        type: 'string',
+                        title: 'Output Collection Property',
+                        description: 'Name of the output property for mapped results',
+                        default: 'results'
+                    },
+                    map_flow_ref: {
+                        type: 'string',
+                        title: 'Map Flow Reference',
+                        description: 'Reference to the flow to execute per item ($component_ref:flow_name)',
+                        default: ''
+                    }
+                }
+            },
+            formats: ['agentspec'],
+            extensions: { 'agentspec:component_type': 'map_node' }
+        }
+    });
+    return registry;
+}
+// ============================================================================
+// Public API
+// ============================================================================
+/** Singleton registry instance */
+const registry = buildRegistry();
+/**
+ * Get FlowDrop NodeMetadata for an Agent Spec component type.
+ *
+ * @param componentType - The Agent Spec component_type value
+ * @returns NodeMetadata for the component type, or undefined if not found
+ *
+ * @example
+ * ```typescript
+ * const metadata = getAgentSpecNodeMetadata('llm_node');
+ * // Returns NodeMetadata with id='agentspec.llm_node', type='default', category='ai', etc.
+ * ```
+ */
+export function getAgentSpecNodeMetadata(componentType) {
+    return registry.get(componentType)?.metadata;
+}
+/**
+ * Get all default Agent Spec node types as FlowDrop NodeMetadata.
+ * These are starter templates — users can provide their own node types instead.
+ *
+ * @returns Array of NodeMetadata for all 9 default Agent Spec node types
+ */
+export function getDefaultAgentSpecNodeTypes() {
+    return Array.from(registry.values()).map((entry) => entry.metadata);
+}
+/**
+ * Get a copy of the NodeMetadata for a component type with custom inputs/outputs.
+ * Used during import to create node metadata that matches the Agent Spec definition's
+ * actual ports rather than the defaults.
+ *
+ * @param componentType - The Agent Spec component_type value
+ * @param inputs - Custom input ports
+ * @param outputs - Custom output ports
+ * @returns A new NodeMetadata with the custom ports merged in
+ */
+export function createAgentSpecNodeMetadata(componentType, inputs, outputs) {
+    const base = registry.get(componentType)?.metadata;
+    if (!base)
+        return undefined;
+    return {
+        ...base,
+        inputs: inputs ?? base.inputs,
+        outputs: outputs ?? base.outputs
+    };
+}

package/dist/adapters/agentspec/index.d.ts

@@ -0,0 +1,37 @@
+/**
+ * Agent Spec Adapter — Barrel Exports
+ *
+ * Provides bidirectional conversion between FlowDrop and Oracle's Open Agent Spec.
+ *
+ * @example
+ * ```typescript
+ * import {
+ *   AgentSpecAdapter,
+ *   AgentSpecAgentAdapter,
+ *   getAgentSpecNodeMetadata,
+ *   validateForAgentSpecExport,
+ *   validateAgentSpecFlow
+ * } from '@d34dman/flowdrop/core';
+ *
+ * // Export a FlowDrop workflow as Agent Spec JSON
+ * const adapter = new AgentSpecAdapter();
+ * const agentSpecJson = adapter.exportJSON(workflow);
+ *
+ * // Import an Agent Spec flow into FlowDrop
+ * const flowDropWorkflow = adapter.importJSON(agentSpecJson);
+ *
+ * // Validate before export
+ * const result = validateForAgentSpecExport(workflow);
+ * if (!result.valid) console.error(result.errors);
+ * ```
+ */
+export { AgentSpecAdapter } from './AgentSpecAdapter.js';
+export { AgentSpecAgentAdapter } from './agentAdapter.js';
+export type { AgentConfig, AgentSpecImportResult } from './agentAdapter.js';
+export { getAgentSpecNodeMetadata, getDefaultAgentSpecNodeTypes, createAgentSpecNodeMetadata } from './defaultNodeTypes.js';
+export { getComponentTypeDefaults, extractComponentType, isAgentSpecNodeId, AGENTSPEC_NAMESPACE } from './componentTypeDefaults.js';
+export type { ComponentTypeDefaults } from './componentTypeDefaults.js';
+export { validateForAgentSpecExport, validateAgentSpecFlow } from './validator.js';
+export type { AgentSpecValidationResult } from './validator.js';
+export { computeAutoLayout } from './autoLayout.js';
+export type { AutoLayoutConfig } from './autoLayout.js';

package/dist/adapters/agentspec/index.js

@@ -0,0 +1,39 @@
+/**
+ * Agent Spec Adapter — Barrel Exports
+ *
+ * Provides bidirectional conversion between FlowDrop and Oracle's Open Agent Spec.
+ *
+ * @example
+ * ```typescript
+ * import {
+ *   AgentSpecAdapter,
+ *   AgentSpecAgentAdapter,
+ *   getAgentSpecNodeMetadata,
+ *   validateForAgentSpecExport,
+ *   validateAgentSpecFlow
+ * } from '@d34dman/flowdrop/core';
+ *
+ * // Export a FlowDrop workflow as Agent Spec JSON
+ * const adapter = new AgentSpecAdapter();
+ * const agentSpecJson = adapter.exportJSON(workflow);
+ *
+ * // Import an Agent Spec flow into FlowDrop
+ * const flowDropWorkflow = adapter.importJSON(agentSpecJson);
+ *
+ * // Validate before export
+ * const result = validateForAgentSpecExport(workflow);
+ * if (!result.valid) console.error(result.errors);
+ * ```
+ */
+// Core adapter
+export { AgentSpecAdapter } from './AgentSpecAdapter.js';
+// Agent-level adapter (wraps flow with agent/tools/LLM config)
+export { AgentSpecAgentAdapter } from './agentAdapter.js';
+// Default node types (optional starter templates — users can provide their own)
+export { getAgentSpecNodeMetadata, getDefaultAgentSpecNodeTypes, createAgentSpecNodeMetadata } from './defaultNodeTypes.js';
+// Component type defaults (adapter infrastructure)
+export { getComponentTypeDefaults, extractComponentType, isAgentSpecNodeId, AGENTSPEC_NAMESPACE } from './componentTypeDefaults.js';
+// Validation
+export { validateForAgentSpecExport, validateAgentSpecFlow } from './validator.js';
+// Auto-layout
+export { computeAutoLayout } from './autoLayout.js';

package/dist/adapters/agentspec/validator.d.ts

@@ -0,0 +1,34 @@
+/**
+ * Agent Spec Validator
+ *
+ * Validates workflows against Agent Spec constraints for export,
+ * and validates imported Agent Spec documents for correctness.
+ */
+import type { AgentSpecFlow } from '../../types/agentspec.js';
+import type { StandardWorkflow } from '../WorkflowAdapter.js';
+/** Validation result */
+export interface AgentSpecValidationResult {
+    valid: boolean;
+    errors: string[];
+    warnings: string[];
+}
+/**
+ * Validate a FlowDrop StandardWorkflow for Agent Spec export compatibility.
+ *
+ * Checks:
+ * - Must have exactly 1 start node (terminal/triggers)
+ * - Must have at least 1 end node (terminal/outputs)
+ * - Gateway nodes must have branches defined
+ */
+export declare function validateForAgentSpecExport(workflow: StandardWorkflow): AgentSpecValidationResult;
+/**
+ * Validate an imported Agent Spec Flow document.
+ *
+ * Checks:
+ * - Has a start_node reference that exists
+ * - At least one end_node exists
+ * - All edge references point to existing nodes
+ * - BranchingNode has branches
+ * - Data flow edges reference valid properties
+ */
+export declare function validateAgentSpecFlow(flow: AgentSpecFlow): AgentSpecValidationResult;