@agentskit/core — for agents

Zero-dependency foundation. Contracts, chat controller, primitives, and a dozen feature subpaths.

#Purpose

Stable TypeScript contracts (Adapter, Tool, Memory, Retriever, Skill, Runtime) + createChatController + primitives. Target: <10 KB gzipped, zero runtime deps.

#Install

npm install @agentskit/core

#Primary exports (main entry)

createChatController(config) — state machine behind every UI binding. Returns { getState, subscribe, send, stop, retry, edit, regenerate, setInput, clear, approve, deny, updateConfig }.
defineTool(config) — typed tool factory with JSON-Schema input inference.
createInMemoryMemory / createLocalStorageMemory — default ChatMemory implementations.
createStaticRetriever / formatRetrievedDocuments — RAG-less retrieval.
executeToolCall, consumeStream, createEventEmitter, safeParseArgs, generateId, buildMessage — low-level helpers.
AgentsKitError, AdapterError, ToolError, MemoryError, ConfigError, RuntimeError, SandboxError, SkillError, ErrorCodes — error taxonomy.
Token budget: compileBudget, approximateCounter.
Progressive tool args: createProgressiveArgParser, executeToolProgressively.
createVirtualizedMemory — hot-window + cold-retriever wrapper (also re-exported from @agentskit/memory).
Multi-modal content parts: textPart, imagePart, audioPart, videoPart, filePart, partsToText, normalizeContent, filterParts.
Agent-loop internals (advanced): buildToolMap, activateSkills, executeSafeTool, createToolLifecycle.
Memory serialization: serializeMessages, deserializeMessages.

#Subpath exports (zero main-bundle weight)

Subpath	Purpose	Recipe
`@agentskit/core/agent-schema`	Declarative agent YAML/JSON + validator	Schema-first agents
`@agentskit/core/prompt-experiments`	A/B prompts with feature flags	A/B prompts
`@agentskit/core/auto-summarize`	Auto-summarizing `ChatMemory` wrapper	Auto-summarize
`@agentskit/core/hitl`	Approval gates + `ApprovalStore`	HITL approvals
`@agentskit/core/security`	PII redactor + injection detector + rate limiter	PII, Injection, Rate limit
`@agentskit/core/compose-tool`	Chain N tools into one	Tool composer
`@agentskit/core/self-debug`	Retry tools with LLM-corrected args	Self-debug
`@agentskit/core/generative-ui`	Typed UI element tree + artifacts	Gen UI
`@agentskit/core/a2a`	Agent-to-Agent protocol spec	Open specs
`@agentskit/core/manifest`	Skill + tool manifest format (MCP-compat)	Open specs
`@agentskit/core/eval-format`	Portable eval dataset + run-result	Open specs

#Minimal example

import { createChatController } from '@agentskit/core'
import { anthropic } from '@agentskit/adapters'

const controller = createChatController({
  adapter: anthropic({ apiKey: process.env.ANTHROPIC_API_KEY!, model: 'claude-sonnet-4-6' }),
})

controller.subscribe(() => console.log(controller.getState().messages))
await controller.send('Hello')

#Common patterns

Wire a controller into any framework via getState + subscribe (see the framework bindings under For agents).
Compose tools with composeTool or wrap failing ones with wrapToolWithSelfDebug.
Gate risky operations with createApprovalGate + HITL + signed audit log.

#Source

Explore nearby

✎ Edit this page on GitHub·Found a problem? Open an issue →·How to contribute →