secondsky/sap-ai-core
|
npx skills add https://github.com/secondsky/sap-skills --skill sap-ai-coreUse this skill when provisioning SAP AI Core, using SAP AI Launchpad, configuring Generative AI Hub orchestration, choosing model providers, building RAG or grounding flows, managing prompt templates, deploying training/inference workloads, or wiring AI capabilities into SAP applications.
10. Structured Output
11. Embeddings
12. ML Training
13. Deployments
14. Bundled Resources
15. SAP AI Launchpad
16. Prompt Registry
17. API Reference
18. Common Patterns
19. Troubleshooting
20. References
SAP AI Core is a service on SAP Business Technology Platform (BTP) that manages AI asset execution in a standardized, scalable, hyperscaler-agnostic manner. SAP AI Launchpad provides the management UI for AI runtimes including the Generative AI Hub.
| Capability | Description |
|------------|-------------|
| Generative AI Hub | Access to LLMs from multiple providers with unified API |
| Orchestration | Modular pipeline for templating, filtering, grounding, masking |
| ML Training | Argo Workflows-based batch pipelines for model training |
| Inference Serving | Deploy models as HTTPS endpoints for predictions |
| Grounding/RAG | Vector database integration for contextual AI |
# Set environment variables from service key
export AI_API_URL="<your-ai-api-url>"
export AUTH_URL="<your-auth-url>"
export CLIENT_ID="<your-client-id>"
export CLIENT_SECRET="<your-client-secret>"
# Get OAuth token
AUTH_TOKEN=$(curl -s -X POST "$AUTH_URL/oauth/token" \
-H "Content-Type: application/x-www-form-urlencoded" \
-d "grant_type=client_credentials&client_id=$CLIENT_ID&client_secret=$CLIENT_SECRET" \
| jq -r '.access_token')
# Check for existing orchestration deployment
curl -X GET "$AI_API_URL/v2/lm/deployments" \
-H "Authorization: Bearer $AUTH_TOKEN" \
-H "AI-Resource-Group: default" \
-H "Content-Type: application/json"
# Create orchestration deployment if needed
curl -X POST "$AI_API_URL/v2/lm/deployments" \
-H "Authorization: Bearer $AUTH_TOKEN" \
-H "AI-Resource-Group: default" \
-H "Content-Type: application/json" \
-d '{
"configurationId": "<orchestration-config-id>"
}'
ORCHESTRATION_URL="<deployment-url>"
curl -X POST "$ORCHESTRATION_URL/v2/completion" \
-H "Authorization: Bearer $AUTH_TOKEN" \
-H "AI-Resource-Group: default" \
-H "Content-Type: application/json" \
-d '{
"config": {
"module_configurations": {
"llm_module_config": {
"model_name": "gpt-4o",
"model_version": "latest",
"model_params": {
"max_tokens": 1000,
"temperature": 0.7
}
},
"templating_module_config": {
"template": [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "{{?user_query}}"}
]
}
}
},
"input_params": {
"user_query": "What is SAP AI Core?"
}
}'
| Plan | Cost | GenAI Hub | Support | Resource Groups |
|------|------|-----------|---------|-----------------|
| Free | Free | No | Community only | Default only |
| Standard | Per resource + baseline | No | Full SLA | Multiple |
| Extended | Per resource + tokens | Yes | Full SLA | Multiple |
Key Restrictions:
SAP AI Core provides access to model providers through a tenant-specific catalog. Treat exact model names and versions as examples until verified in the target tenant with GET /v2/lm/scenarios/foundation-models/models or SAP AI Launchpad Model Library.
For detailed provider configurations and model lists, see references/model-providers.md.
The orchestration service provides unified access to multiple models through a modular pipeline with 8 execution stages:
For complete orchestration module configurations, examples, and advanced patterns, see references/orchestration-modules.md.
Azure Content Safety: Filters content across 4 categories (Hate, Violence, Sexual, SelfHarm) with severity levels 0-6. Azure OpenAI blocks severity 4+ automatically. Additional features include PromptShield and Protected Material detection.
Llama Guard 3: Covers 14 categories including violent crimes, privacy violations, and code interpreter abuse.
Two PII protection methods:
MASKED_ENTITY (non-reversible)MASKED_ENTITY_ID (reversible)Supported entities (25 total): Personal data, IDs, financial information, SAP-specific IDs, and sensitive attributes. For complete entity list and implementation details, see references/orchestration-modules.md.
Integrate external data from SharePoint, S3, SFTP, SAP Build Work Zone, and DMS. Supports PDF, HTML, DOCX, images, and more. Limit: 2,000 documents per pipeline with daily refresh. For detailed setup, see references/grounding-rag.md.
Enable LLMs to execute functions through a 5-step workflow: define tools → receive tool_calls → execute functions → return results → LLM incorporates responses. Templates available in templates/tool-definition.json.
Force model responses to match JSON schemas using strict validation. Useful for structured data extraction and API responses.
Generate semantic embeddings for RAG and similarity search via /v2/embeddings endpoint. Supports document, query, and text input types.
Uses Argo Workflows for training pipelines. Key requirements: create default object store secret, define workflow template, create configuration with parameters, and execute training. For complete workflow patterns, see references/ml-operations.md.
Deploy models via two-step process: create configuration (with model binding), then create deployment with TTL. Statuses: Pending → Running → Stopping → Stopped/Dead. Templates in templates/deployment-config.json.
Web-based UI with 4 key applications:
Required roles include genai_manager, genai_experimenter, prompt_manager, orchestration_executor, and mloperations_editor. For complete guide, see references/ai-launchpad-guide.md.
The Prompt Registry manages the lifecycle of prompt templates from design to runtime, integrating them into SAP AI Core and orchestration workflows.
Two management interfaces:
Key endpoints:
POST /v2/lm/promptTemplates — Create a prompt templatePOST /v2/lm/promptTemplates/{id}/substitution — Fill template by IDPOST /v2/lm/scenarios/{scenario}/promptTemplates/{name}/versions/{version}/substitution — Fill by nameFor complete Prompt Registry documentation, see references/ai-launchpad-guide.md.
Key endpoints: /v2/lm/scenarios, /v2/lm/configurations, /v2/lm/deployments, /v2/lm/executions, /lm/meta. For complete API reference with examples, see references/api-reference.md.
CAP Integration: SAP CAP is the primary consumer framework for AI Core on BTP. Bind an AI Core service instance to your CAP app via MTA, then call the orchestration API from CAP event handlers using the SAP Cloud SDK for AI. Always process LLM calls asynchronously in production (return 202 Accepted, process in background via cds.spawn) to avoid BTP load balancer timeouts. See sap-cap-capire and sap-cloud-sdk-ai skills for complete code examples.
Simple Chat: Basic model invocation with templating module
RAG with Grounding: Combine vector search with LLM for context-aware responses
Secure Enterprise Chat: Filtering + masking + grounding for PII protection
Templates available in templates/orchestration-workflow.json.
Common Issues:
default object store secretRequest quota increases via support ticket (Component: CA-ML-AIC).
references/orchestration-modules.md - All orchestration modules in detailreferences/generative-ai-hub.md - Complete GenAI hub documentationreferences/model-providers.md - Model providers and configurationsreferences/api-reference.md - Complete API endpoint referencereferences/grounding-rag.md - Grounding and RAG implementationreferences/ml-operations.md - ML operations and trainingreferences/advanced-features.md - Chat, applications, security, auditingreferences/ai-launchpad-guide.md - Complete SAP AI Launchpad UI guidetemplates/deployment-config.json - Deployment configuration templatetemplates/orchestration-workflow.json - Orchestration workflow templatetemplates/tool-definition.json - Tool calling definition templateTake secondsky/sap-ai-core from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.