Common Operators Catalog

bloge-common-operators provides ready-to-use operators across integration, AI, data shaping, notification, approval, storage, crypto, validation, and dynamic graph execution. The module is designed as an operator catalog, not as a mandatory dependency of the core runtime.

Registration Pattern

var registry = new DefaultOperatorRegistry();

CommonOperators.builder()
    .http()
    .transform()
    .validation()
    .crypto(secretProvider)
    .ai(llmProvider)
    .dynamic()
    .build()
    .registerAll(registry);

Provider-backed categories require the corresponding SPI implementation. Self-contained categories can be registered without external services.

Categories

Category	Operators	Required provider
HTTP	httpRequest, graphqlQuery, webhookTrigger, httpPoll	JDK HTTP client plus optional verifier
Database	sqlQuery, sqlExecute, sqlBatch, noSqlQuery, redis	DataSourceProvider, Redis/NoSQL providers where applicable
Messaging	messagePublish, messageConsume, eventEmit	MessageBrokerProvider
AI / LLM	llmChat, llmStreamingChat, embedding, textClassify, textSummarize, ragRetrieve, structuredExtract, textSplit, conversation	LlmProvider, EmbeddingProvider, VectorStoreProvider, optional TokenEstimator
Transform	jsonParse, jsonStringify, xmlParse, csvParse, templateRender, regexExtract, dataMapping	None
Notification	emailSend, smsSend, multiChannelNotify	Mail and SMS providers
Approval	approvalRequest, approvalCheck	ApprovalService
Storage	fileRead, fileWrite, objectStorage	ObjectStorageProvider for object storage
Crypto	hash, encryptDecrypt, jwt	SecretProvider for secret-backed operations
Validation	schemaValidate, ruleEngine	None
Dynamic	dynamicSubGraph	Current OperatorRegistry and DSL sandbox policy

AI and Conversation Operators

The AI operator set is the runtime foundation for agent workflows, RAG pipelines, and structured model calls.

Operator	Use case
llmChat	Text or structured chat completion
llmStreamingChat	Chunked model output
conversation	Conversation memory with full, sliding-window, or token-budget strategies
ragRetrieve	Retrieval over a vector store
structuredExtract	Schema-shaped extraction
textClassify / textSummarize / textSplit	Common text processing primitives

Agent loops in bloge-agent-ext build on these provider contracts but remain a separate extension runtime. See Agents.

Dynamic Subgraphs

dynamicSubGraph validates runtime-generated BLOGE DSL through a sandbox policy, compiles it with the current operator registry, and executes it as a nested sub-graph.

Use it when:

• an upstream step generates a small workflow
• generated DSL must be sandboxed before execution
• the resulting graph should still use normal BLOGE scheduling, resilience, and observability

Do not use it as a substitute for ordinary imports or checked-in DSL. Runtime graph generation needs strict policy, size limits, and audit visibility.

Design Rules

• Register only the operator categories the service actually uses.
• Keep provider-backed operators behind clear service abstractions.
• Treat operator input and output schemas as public contracts for lint, editor tooling, and agents.
• Prefer common operators for commodity integration work; write domain operators for business decisions and side effects.
• Add explicit timeout and fallback policies around operators that call external systems.

Next Steps

• Learn schema contracts in Schema & I/O.
• Design custom operators with Design Principles.
• Use AI operators with AI Agents & LLM Operators.