cloud-ide-ai

Angular library for CloudIDE AI features. It provides the UI-facing API and components that talk to the backend AI stack (policy, retrieval, model router, audit).

Workspace AI documentation

Backend behaviour, policy, and rollout are defined in the Node suite docs. Always refer to these when integrating or changing AI behaviour:

• cloudide-node-suite/docs/ai-architecture-and-policy.md – Request flow, capability guardrails (Phase-1 read-only), policy principles, reliability controls.
• cloudide-node-suite/docs/ai-rollout-playbook.md – Stages (internal → pilot → GA), rollback procedure.
• cloudide-node-suite/docs/ai-observability-security-checklist.md – Runtime/rollout/security controls, monitoring, production hardening.

Public API

• AiApiService – Use this as the single entry for all AI calls. Methods:
  • chat(query, context, capabilityCode?) – Basic chat (default ai_basic_chat).
  • pageSummary(query, context) – Page summary.
  • recordExplain(query, context, recordId?) – Explain a record.
  • recommendations(query, context) – Recommendations (read-only).
  • getCapabilities() – Allowed capabilities for current user.
  • feedback(requestId, rating, feedbackText?) – Send feedback for a request.
• AiContextService – Current route context (pageCode, module, url) for AI requests. Required for policy (e.g. sypg_page_code).
• AI_CAPABILITY_CODES, AI_PHASE1_CAPABILITIES – Constants aligned with backend Phase-1 capabilities.

Usage example

import { AiApiService, AiContextService, AI_CAPABILITY_CODES } from 'cloud-ide-ai';

// Inject AiApiService and AiContextService
const ctx = this.contextService.getCurrentContext();
this.aiApiService.chat('What is on this page?', ctx, AI_CAPABILITY_CODES.BASIC_CHAT).subscribe({
  next: (res) => { /* res.data has message, citations, policy, provider, latency_ms */ },
  error: (err) => { /* handle */ }
});

Routes and components

• Routes – ai/chat, ai/assist (see cloud-ide-ai.routes.ts). Both use the same chat page with different sypg_page_code.
• Components – AiChatPageComponent, AiDrawerComponent for full-page and drawer UX.

Backend requirements

• Backend must expose AI routes under the base URL used by hostManagerRoutesUrl.cideSuiteHost and aiRoutesUrl from cloud-ide-lms-model (e.g. /ai/chat, /ai/page-summary, /ai/record-explain, /ai/recommendations, /ai/capabilities/me, /ai/feedback).
• Set OPENAI_API_KEY in the Node suite for production LLM responses; otherwise the backend uses a mock response (see cloudide-node-suite/src/services/ai/modelRouterService.ts).