@contractspec/lib.ai-agent 7.0.5 → 7.0.7

package/README.md

@@ -1,137 +1,78 @@
 # @contractspec/lib.ai-agent
 
-[![npm version](https://img.shields.io/npm/v/@contractspec/lib.ai-agent)](https://www.npmjs.com/package/@contractspec/lib.ai-agent)
-[![npm downloads](https://img.shields.io/npm/dt/@contractspec/lib.ai-agent)](https://www.npmjs.com/package/@contractspec/lib.ai-agent)
-[![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/lssm-tech/contractspec)
+Website: https://contractspec.io
 
-Website: https://contractspec.io/
+**Core AI agent runtime for ContractSpec with tool orchestration, MCP integration, session state, memory, and telemetry.**
 
-**AI governance for ContractSpec** — Constrain what AI agents can change, enforce contracts they must respect.
+## What It Provides
 
-Stateful AI agent orchestration with type-safe specs, tool execution, knowledge bindings, and per-tenant guardrails. Agents read contracts as their source of truth, enabling triage, growth experiments, and DevOps automation with contract-enforced safety.
+- Provides the central agent runtime used by chat, automation, and higher-level orchestration packages.
+- Supports tools, sessions, memory, approvals, providers, telemetry, and MCP-aware workflows.
+- Acts as the stable public API for agent execution across multiple runtimes and delivery surfaces.
+- `src/providers/` contains provider integrations and provider-facing adapters.
 
-## Features
+## Installation
 
-- Type-safe `AgentSpec` + registry for declarative agent definitions
-- Tool loop orchestration with runtime `maxSteps` overrides and confidence-aware escalation hooks
-- Memory framework that mixes working memory with long-term persistence hooks
-- Tool registry/executor with structured input validation, timeout, and cooldown controls
-- Approval workflow helpers for human-in-the-loop gates (see `/approval`)
-- MCP client support for `stdio`, `sse`, and `http` transports
-- Adapter-first runtime interop surfaces for LangGraph, LangChain, and workflow-devkit style integrations
-
-## Runtime adapter surfaces
-
-`AgentSpec.runtime` and `interop/runtime-adapters` expose optional ports for:
-
-- checkpoint storage
-- suspend/resume orchestration
-- retry classification
-- approval gateways
-
-These ports are optional by design and keep external runtime dependencies decoupled from core agent contracts.
-
-## Operations as Tools
-
-**Preferred pattern:** One contract → multiple surfaces. Reference a ContractSpec operation in your agent; the tool handler and input schema are derived automatically.
-
-```ts
-import { defineAgent } from '@contractspec/lib.ai-agent';
-import { defineQuery } from '@contractspec/lib.contracts-spec';
-import { OperationSpecRegistry } from '@contractspec/lib.contracts-spec';
-import { createAgentFactory } from '@contractspec/lib.ai-agent/agent/agent-factory';
-
-// 1. Define operation (existing pattern)
-const SearchKnowledgeQuery = defineQuery({
-  meta: { key: 'knowledge.search', version: '1.0.0', owners: ['@platform'], goal: '...', context: '...' },
-  io: { input: SearchInputModel, output: SearchOutputModel },
-});
-
-// 2. Register and bind handler
-const ops = new OperationSpecRegistry().register(SearchKnowledgeQuery).bind(SearchKnowledgeQuery, handler);
-
-// 3. Reference in agent
-const agent = defineAgent({
-  meta: { key: 'support.bot', version: '1.0.0', owners: ['@platform'] },
-  instructions: 'Resolve support tickets.',
-  tools: [
-    { name: 'search_knowledge', operationRef: { key: 'knowledge.search', version: '1.0.0' }, automationSafe: true },
-  ],
-});
-
-// 4. Create factory with operationRegistry
-const factory = createAgentFactory({
-  defaultModel,
-  registry: agentRegistry,
-  toolHandlers: new Map(),
-  operationRegistry: ops,
-});
-```
-
-**Output rendering:** When tool output should render via PresentationSpec, FormSpec, or DataViewSpec, add `outputPresentation`, `outputForm`, or `outputDataView` to `AgentToolConfig` (at most one per tool). The tool adapter wraps raw output as `{ presentationKey, data }`, `{ formKey, defaultValues }`, or `{ dataViewKey, items }` for `ToolResultRenderer`. OperationSpec can also declare these refs; when the tool has no output refs, the operation's refs are used as fallback.
-
-**Fallback (inline tools):** When the tool is not an operation (LLM subcalls, external APIs), use inline `AgentToolConfig` with `schema` and a manual handler in `toolHandlers`.
-
-## Bundle spec / surface-runtime integration
-
-When building planner agents for `@contractspec/lib.surface-runtime`, planner tools (e.g. `propose-patch`) from `@contractspec/lib.surface-runtime/runtime/planner-tools` map to `AgentToolConfig`. Wire `proposePatchToolConfig` into `AgentSpec.tools`; the handler should validate via `validatePatchProposal` and return `SurfacePatchProposal`. Tools from surface-runtime planner-tools are `AgentToolConfig`-compatible.
-
-## Quickstart
-
-```ts
-import { AgentRegistry, AgentRunner, defineAgent } from '@contractspec/lib.ai-agent';
-import { ToolExecutor } from '@contractspec/lib.ai-agent/tools';
-import { InMemoryAgentMemory } from '@contractspec/lib.ai-agent/memory';
-
-const SupportAgent = defineAgent({
-  meta: { name: 'support.bot', version: '1.0.0', owners: ['team-support'], domain: 'operations' },
-  instructions: 'Resolve support tickets. Escalate whenever confidence < 0.75.',
-  tools: [{ name: 'search_knowledge', description: 'Query support corpus', schema: { type: 'object', properties: { question: { type: 'string' } }, required: ['question'] } }],
-  policy: { confidence: { min: 0.75 }, escalation: { auto: true } },
-});
-
-const registry = new AgentRegistry().register(SupportAgent);
-const runner = new AgentRunner({
-  registry,
-  llm: mistralProvider,
-  toolExecutor: new ToolExecutor({ tools: [searchKnowledgeTool] }),
-  memoryManager: new InMemoryAgentMemory(),
-});
-
-const result = await runner.run({ agent: 'support.bot', input: 'My payout failed.', tenantId: 'acme' });
-if (result.requiresEscalation) notifyHuman(result);
-```
-
-See `examples/ai-support-bot` for a full workflow including ticket ingestion and approval queues.
-
-## MCP Client Tooling
-
-```ts
-import { createAgentFactory } from '@contractspec/lib.ai-agent/agent/agent-factory';
-
-const factory = createAgentFactory({
-  defaultModel,
-  registry,
-  toolHandlers,
-  mcpServers: [
-    {
-      name: 'filesystem',
-      command: 'npx',
-      args: ['-y', '@modelcontextprotocol/server-filesystem', '/workspace'],
-      toolPrefix: 'fs',
-    },
-    {
-      name: 'posthog',
-      transport: 'http',
-      url: process.env.POSTHOG_MCP_URL!,
-      accessTokenEnvVar: 'POSTHOG_MCP_TOKEN',
-      toolPrefix: 'ph',
-    },
-  ],
-});
-
-const agent = await factory.create('support.bot');
-const result = await agent.generate({ prompt: 'Summarize today\'s incidents.' });
-
-await agent.cleanup();
-```
+`npm install @contractspec/lib.ai-agent`
+
+or
+
+`bun add @contractspec/lib.ai-agent`
+
+## Usage
+
+Import the root entrypoint from `@contractspec/lib.ai-agent`, or choose a documented subpath when you only need one part of the package surface.
+
+## Architecture
+
+- `src/agent/`, `src/session/`, and `src/memory/` contain the runtime core for execution state and persistence hooks.
+- `src/tools/`, `src.providers/`, and `src.interop/` connect providers, tools, and MCP-aware runtime surfaces.
+- `src.telemetry/`, `src.approval/`, `src.knowledge/`, `src.schema/`, and `src.spec/` round out agent policy and observability surfaces.
+- `src/index.ts` is the root public barrel and package entrypoint.
+- `src/types.ts` is shared public type definitions.
+
+## Public Entry Points
+
+- Large multi-subpath library exporting agent runtime, approval, memory, knowledge, providers, schema, telemetry, tools, and types.
+- Export `.` resolves through `./src/index.ts`.
+- Export `./agent` resolves through `./src/agent/index.ts`.
+- Export `./agent/agent-factory` resolves through `./src/agent/agent-factory.ts`.
+- Export `./agent/contract-spec-agent` resolves through `./src/agent/contract-spec-agent.ts`.
+- Export `./agent/json-runner` resolves through `./src/agent/json-runner.ts`.
+- Export `./agent/unified-agent` resolves through `./src/agent/unified-agent.ts`.
+- Export `./approval` resolves through `./src/approval/index.ts`.
+- Export `./approval/workflow` resolves through `./src/approval/workflow.ts`.
+- Export `./exporters` resolves through `./src/exporters/index.ts`.
+- Export `./exporters/claude-agent-exporter` resolves through `./src/exporters/claude-agent-exporter.ts`.
+- The package publishes 66 total export subpaths; keep docs aligned with `package.json`.
+
+## Local Commands
+
+- `bun run dev` — contractspec-bun-build dev
+- `bun run build` — bun run prebuild && bun run build:bundle && bun run build:types
+- `bun run test` — bun test
+- `bun run lint` — bun lint:fix
+- `bun run lint:check` — biome check .
+- `bun run lint:fix` — biome check --write --unsafe --only=nursery/useSortedClasses . && biome check --write .
+- `bun run typecheck` — tsc --noEmit
+- `bun run publish:pkg` — bun publish --tolerate-republish --ignore-scripts --verbose
+- `bun run publish:pkg:canary` — bun publish:pkg --tag canary
+- `bun run clean` — rimraf dist .turbo
+- `bun run build:bundle` — contractspec-bun-build transpile
+- `bun run build:types` — contractspec-bun-build types
+- `bun run prebuild` — contractspec-bun-build prebuild
+
+## Recent Updates
+
+- Missing dependencies (thanks to knip).
+- Replace eslint+prettier by biomejs to optimize speed.
+- Agentic workflows — subagents, memory tools, and next steps.
+- Vnext ai-native.
+- Backend operations + frontend rendering support.
+- Add latest models and align defaults.
+
+## Notes
+
+- High blast radius — used by multiple bundles and libs.
+- `AgentSpec` interface is a public contract; do not change its shape without a migration plan.
+- MCP transport adapters must stay runtime-agnostic (no Node/browser-specific globals).

package/dist/agent/agent-factory.d.ts

@@ -1,16 +1,16 @@
-import type { LanguageModel, Tool } from 'ai';
-import type { KnowledgeRetriever } from '@contractspec/lib.knowledge/retriever';
 import type { ModelSelector } from '@contractspec/lib.ai-providers/selector-types';
 import type { OperationSpecRegistry } from '@contractspec/lib.contracts-spec/operations/registry';
-import type { AgentSpec } from '../spec/spec';
-import type { AgentRegistry } from '../spec/registry';
-import type { ToolHandler } from '../types';
-import type { McpClientConfig } from '../tools/mcp-client';
-import type { SubagentRegistry } from '../tools/tool-adapter';
-import type { AgentMemoryStore } from '../tools/agent-memory-store';
+import type { KnowledgeRetriever } from '@contractspec/lib.knowledge/retriever';
+import type { LanguageModel, Tool } from 'ai';
 import type { AgentSessionStore } from '../session/store';
+import type { AgentRegistry } from '../spec/registry';
+import type { AgentSpec } from '../spec/spec';
 import type { TelemetryCollector } from '../telemetry/adapter';
 import type { PostHogLLMConfig, PostHogTracingOptions } from '../telemetry/posthog-types';
+import type { AgentMemoryStore } from '../tools/agent-memory-store';
+import type { McpClientConfig } from '../tools/mcp-client';
+import type { SubagentRegistry } from '../tools/tool-adapter';
+import type { ToolHandler } from '../types';
 import { ContractSpecAgent } from './contract-spec-agent';
 /**
  * Factory configuration for creating agents.