BrowseAll ecosystems
Everything in the atlas
198 of 2341
Type
Domain: AI & Agents
Origin
Signals: Last 3 months
Content
More
Group by: Capability
EntryBySignalsLast activity
@cap-js/agents@cap-js/agentsSAP's own CDS plugin for building agents inside a CAP service, wired to audit logging and telemetry through peer dependencies rather than leaving observability to the author.sap official, cap
CDS plugin for building agents inside a CAP service
Not as a gateway or proxy for SAP Application APIs: capire's guide states that CAP-level agents are intended only for exposing custom CAP application services and are not an SAP-endorsed architecture or pathway for agentic access to those APIs, and sends the reader to the SAP Architecture Center for the endorsed routes. Not as the screen a user of the application talks to, either — the chat preview in the generic index.html page is marked experimental and for local development.
@mi8y/cap-agents-memory@mi8y/cap-agents-memoryA LangGraph Store written against CDS: the facts and preferences an agent keeps between sessions are entities in the CAP database, read and written inside a request handler under its tenant.independent, cap
LangGraph Store keeping agent memory as CDS entities
@mi8y/cap-agents-sdm-backend@mi8y/cap-agents-sdm-backendA Deep Agents filesystem backend whose files are documents in an SAP Document Management repository, so what a CAP agent reads and writes stays where people can open it too.independent, cap
Deep Agents filesystem over an SAP Document Management repository
Not a way to edit Office formats: the author states that access to a file does not automatically provide native editing of Word documents, spreadsheets or PDFs, and suggests starting with text-based content such as instructions, templates and reports. Not a control point either: a prompt telling the agent to ask for confirmation or to leave a template alone needs enforceable application controls where that is a business requirement, and the host application stays responsible for the authenticated request context and tenant isolation.
@mi8y/cds-langgraph-persistence@mi8y/cds-langgraph-persistenceA LangGraph checkpointer backed by CDS, so an agent's state survives the request that created it — durability for LangChain and LangGraph agents running inside the CAP Node.js runtime.independent, cap
LangGraph checkpointer backed by CDS for CAP agents
@sap/joule-studio-cli@sap/joule-studio-cliSAP’s CLI package for both legacy Joule digital assistants and the new Joule Studio solution workflow, through the joule and jl commands.sap official, btp
CLI for Joule digital assistants and Joule Studio solutions
@sap/joule-work-dev-cli-early-adopters@sap/joule-work-dev-cli-early-adoptersjl, the pro-code CLI of the new Joule Studio: a solution is a folder with a solution.yaml and typed assets, validated, built into a zip and deployed to Joule Studio's managed runtime, with Claude Code or OpenCode wired in over MCP. Early Adopter Care.sap official, btp
Pro-code CLI for building and deploying Joule Studio solutions
@saptarishi/cds-plugin-llm@saptarishi/cds-plugin-llmA CAP service kind that puts eleven model providers behind one chat, stream, embed and batch interface, so which backend a service calls is a profile in package.json rather than a change to the handler.independent, cap
CAP service kind putting eleven model providers behind one API
cap-agent-ui5-webuicap-agent-ui5-webuiGives a CAP agent a UI5 web interface out of the box, so an agent built in the service has somewhere to be talked to without a separate frontend project.independent, cap
UI5 web interface for an agent built in a CAP service
Not where the agent is reached over anything other than A2A: the README states the plugin does not speak anything other than A2A as @cap-js/agents serves it. Not where an upgrade has to be uneventful either — it pins @cap-js/agents to exactly 0.9.1, calls that package pre-1.0 and days old at the time of writing, and says this plugin inherits that instability and to expect breakage on upgrade.
cds-agentscds-agentsTurns a CAP service into a tool layer for agent frameworks — a CAPToolkit bound to a named service, with a tool strategy controlling how much of it becomes callable.independent, cap
Toolkit turning a CAP service into tools for agent frameworks
cap-agent-ui5-webui-samplecap-agent-ui5-webui-sampleA worked example of the cap-agent-ui5-webui plugin, shown running locally, in hybrid mode against SAP AI Core, and deployed.independent, cap
Worked example of the cap-agent-ui5-webui plugin
Neither the package manifest nor the repository tree declares a licence.
SAP-samples/btp-agentic-ai-use-casesSAP-samples/btp-agentic-ai-use-casesTwo runnable custom industry AI agent projects from an SAP webinar series, one built in Joule Studio Agent Builder and one wiring Joule to a code-based deep research agent.sap official, btp
Two runnable industry AI agent projects built around Joule
SAP-samples/codejam-code-based-agentsSAP-samples/codejam-code-based-agentsA CodeJam teaching code-based AI agents in Python or TypeScript, from a first agent through custom tools to deployment on SAP BTP and integration with Joule.sap official, btp
CodeJam on code-based AI agents in Python or TypeScript on SAP BTP
xtravelsxtravelsA travel-booking application built on the master data that xflights publishes, composing another package's schema into its own rather than owning master data itself.sap official, cap
CAP travel-booking app built on xflights master data
Achieve Agentic Enterprise Automation with SAP Integration Suite and Joule StudioAchieve Agentic Enterprise Automation with SAP Integration Suite and Joule StudioA Devtoberfest 2026 integration-track session (INT-22), the second of that week's integration sessions, given by an SAP product expert for process automation and integration together with the product manager for Joule Studio, who walks the audience through it. The opening states the subject: process management, automation and orchestration unified into one combined approach, which the presenters put forward as the foundation for building custom AI agents.SAP Developers, published 2026-09-30
Adapting SAP Joule for Purchase Order Creation from an Existing Purchase RequisitionAdapting SAP Joule for Purchase Order Creation from an Existing Purchase RequisitionA custom Joule Agent and Joule Skill, built in Joule Studio, classic edition to extend SAP Joule past its built-in Purchase Order display and change features: a buyer's single sentence naming a Purchase Requisition number becomes a created Purchase Order, with the skill wrapping the Purchase Requisition and Purchase Order OData V4 APIs as Action Packages reached through two destinations. The author anchors the agent's success condition to a PurchaseOrderNumber in the skill's output rather than an HTTP status code, and reports that the Purchase Requisition item structure does not map one-to-one onto the Purchase Order creation payload.jonathanvi59, published 2026-08-19
Agentic AI & AI AgentsAgentic AI & AI AgentsSAP's target picture for AI agents: Joule as the engagement layer — Joule Work, Joule Assistants, Joule Agents — with custom agents from customers, partners and SAP teams connecting through A2A for agent-to-agent collaboration and MCP for tool connectivity, on the SAP Business AI Platform. It says up front that parts are not generally available yet, in particular bidirectional communication with third-party agents through the Agent Gateway, so it reads as direction with the building blocks named rather than a design to copy today.published 2026-08-27
Be the First to Understand Joule Work's Agentic ThinkingBe the First to Understand Joule Work's Agentic ThinkingA Devtoberfest 2026 session on the agent harness behind agentic thinking in Joule Work: generic operating-system tools in a gVisor-isolated sandbox instead of hand-wired domain functions, a react loop, a warm pool of sandboxes torn down after each task, tenant-separated skill loading, and placeholder tokens the agent never sees swapped for real credentials at the egress path. A skill is a folder with a SKILL.md carrying metadata and instructions, and where it reaches a system, an SAP connectivity JSON that declares each connector by ORD ID with its guardrails — GET allowed by default, POST, PUT, PATCH and DELETE requiring human approval, everything else blocked. The demo uploads a cost-centre skill through the skills view, then has agentic thinking delegate over A2A through Agent Gateway to a custom agent built in Joule Studio, whose MCP tools read cost centres from an S/4HANA Cloud system, and ends by turning the result into a PDF; the Joule Studio trace is what shows the tool call actually happened.SAP Developers, published 2026-10-15
Build Your AI Agent prototype for free using SAP CAP, LangChain and Ollama cloudBuild Your AI Agent prototype for free using SAP CAP, LangChain and Ollama cloudA CAP Node.js service action calls an Ollama-hosted model through LangChain's ChatOllama, with the API key read from a .env file instead of any destination or credential binding — the hand-rolled-key route SAP's own SDK call exists to avoid. The same handler then builds a LangChain agent with createAgent and one mock tool, so a question about restaurants triggers a tool call and any other question does not, independent of the @cap-js/agents plugin's A2A-served harness.rajeevgoswami1, published 2026-08-29
Building a Pro-Code Agent on SAP BTP — Part 2: The Agent (LangGraph + GenAI Hub)Building a Pro-Code Agent on SAP BTP — Part 2: The Agent (LangGraph + GenAI Hub)Part 2 of a three-part series builds a separate, non-CAP Node/TypeScript agent that calls the Part 1 CAP service's MCP tools and reasons through SAP AI Core via @sap-ai-sdk/orchestration, using LangGraph's createAgent ReAct loop for provider-agnostic model calls across GPT, Claude and Gemini deployments. It explains why the tool's Zod schema and model-facing description are the agent's actual contract, the gotcha between a foundation-models deployment and an orchestration-scenario one, and how edge-case testing replaced date inference with a server-computed daysOverdue field after the agent wrongly called a delayed order 'on schedule'.rodfior, published 2026-08-20
Building a Pro-Code Agent on SAP BTP — Part 3: Joule Integration via A2ABuilding a Pro-Code Agent on SAP BTP — Part 3: Joule Integration via A2AThe third instalment of a three-part series deploys an already-built CAP service and LangGraph agent to SAP BTP, Cloud Foundry Runtime, wraps the agent in an A2A protocol server, and registers it with SAP Joule through a capability YAML and the @sap/joule-studio-cli, confirming the chain works end to end inside a real Joule assistant. It covers CAP's default JWT/XSUAA authentication in production, what a test deployment needs to get past it, and the cf push, mbt build and environment wiring that run both apps as separate Cloud Foundry containers. A closing section names what stays unfinished by design — a long-lived AI Core service key, no authentication on the BTP destination, no principal propagation — as the gap between this MVP and a production posture.rodfior, published 2026-08-20
Building Agentic Workflows with n8n in Joule StudioBuilding Agentic Workflows with n8n in Joule StudioA Devtoberfest 2026 session, JS-31, on SAP-managed n8n inside Joule Studio, shown as a labs preview on a blocked-invoice scenario: a solution made of assets, an intent refined by the web agent into requirements and then a specification, and the generated workflow opened in the n8n editor that Joule Studio connects to its own n8n tenant, with the generated files committed through GitHub. The nodes carry the other edges — the human-in-the-loop node is built on SAP Task Center, and SAP AI Core is where model providers are configured centrally instead of per developer. It ends in Joule chat, where the approval task the workflow created is listed, opened and completed.SAP Developers, published 2026-10-08
Building Code-Based AI Agents on the BAIPBuilding Code-Based AI Agents on the BAIPA Devtoberfest 2026 session on writing the agent yourself rather than configuring one, which is the agent tooling edge: an agent is a model with tools, skills, memory and context around it, and the session reads the code of one built in CAP as a LangGraph state graph, with an agent node that calls the model, a tools node holding the MCP clients, and the edge between them looping until no tool is needed. It then shows the shorter route, a CDS service annotated for the agent protocol with its instructions and skills in an agent.md file. SAP AI Core carries both, with the CAP service in front of it as middleware so the credentials never reach the browser, and that is what cap-agent-ui5-webui is demonstrated against: added as a dependency, it gives the annotated service a UI5 chat interface that can be deployed rather than only previewed. A Python multi-agent application on Cloud Foundry, where an orchestrator delegates to per-stack specialists so no single model sees every tool, closes the session.published 2026-10-09
CAP in the age of AICAP in the age of AISeven live episodes of Hands-on SAP Dev in which DJ Adams works through CAP's AI features with the people who build them, each one a screen-share against a sample project. The first three are @cap-js/mcp: the @mcp annotation on a service, the describe, query and call tools, CQL as the query format, and what authorisation takes out of the tool list. From episode three the run moves to A2A and @cap-js/agents — human approval of an action, agents.md and skills, session persistence, telemetry and sub-agents — and the last two episodes are vector embeddings in CDS, local on SQLite and native on SAP HANA Cloud.published 2026-08-07
CAP in the age of AI #3 - from MCP to A2ACAP in the age of AI #3 - from MCP to A2AThe episode that finishes @cap-js/mcp with authorisation and then crosses to agents. @requires and @restrict are the same annotations CAP already has, and the plugin reads them to filter the tool list and the describe output per user and per request, so a model is not told about entities the caller cannot reach — where the OData metadata of the same service lists every exposed entity regardless. The second half annotates a service with @agent from @cap-js/agents, which serves it over A2A with an agent card and a chat preview, runs first against a mocked response and then against SAP AI Core bound with cds bind and a hybrid profile, and reuses the MCP adapter's tools rather than defining new ones.published 2026-08-21
CAP in the age of AI #4 - more on agentsCAP in the age of AI #4 - more on agentsAsking a human first, and telling the agent how to behave. Ordering a book through @cap-js/agents happens without a question being asked; the human-in-the-loop annotation on the action turns the same request into an approve-or-reject step, which the episode follows down to the A2A status update that stores the task in an input-required state in the database — A2A itself being JSON-RPC 2.0 over HTTP with server-sent events for the stream. The rest is instructions on disk: an agents.md in the service's own folder removes the question about how many copies to order, a skill under a skills folder is advertised by its description and read only when the agent decides the case applies, and either file makes the plugin build a deep agent instead of the plain one.published 2026-09-11
CAP in the age of AI #5 - agents wrap-upCAP in the age of AI #5 - agents wrap-upThe agents topic closed out on what sits around the agent rather than inside it, now that the documentation has moved into capire. The chat preview's session is persisted in the database rather than kept in memory, traces go out through @cap-js/telemetry to a self-hosted MLflow under the OpenTelemetry conventions for generative AI, and the panel is explicit that @agent reuses the MCP adapter's code programmatically without serving an MCP endpoint. It ends in the xtravels sample, where a travel agent delegates to a hotel-service sub-agent and reaches a separately started flight service over MCP through the service binding CAP resolves anyway. Earlier in the hour the same agent answers from a model running on the presenter's laptop, reached as an OpenAI-compatible endpoint.published 2026-10-02
CAP: The Road Ahead - Simplicity in an AI EraCAP: The Road Ahead - Simplicity in an AI EraDevtoberfest 2026 session CAP10, from the head of SAP's Cloud Application Programming Model: forty minutes of live building rather than slides, on the argument that the cheapest code in an AI project is code nobody has to generate. Claude Code builds an invoice-matching workspace from cds init onwards, and each step is one the framework answers rather than the agent: a 42-line domain model that @sap/cds serves as OData and Fiori elements with a status flow, data privacy and change-tracking annotations that produce audit log entries in the console and would reach the SAP Audit Log service once deployed, an imported S/4HANA supplier data product, and semantic matching from a local embedding model after @cap-js/ai is added. The last three steps are annotations: @mcp beside @odata turns the service into an MCP server that a second Claude session queries through @cap-js/mcp, an agent annotation adds the A2A protocol with an agent card and an audit log entry for the agent task, and a skill file gives the agent the matching policy.SAP Developers, published 2026-09-25
Combining AI Agents with the power of SAP HANA embedded ML (part 1)Combining AI Agents with the power of SAP HANA embedded ML (part 1)Devtoberfest 2026 session DAT-12, the first of two parts, in which Vitaliy Rudnytskiy does a data scientist's work against the embedded machine learning of SAP HANA Cloud by talking to an agent. The session first takes an agent apart into its LLM, its tools, its memory and the loop between them, then stacks the Predictive Analysis Library under the hana-ml Python wrapper and that under the HANA AI toolkit, whose tools an agent calls from plain English in a notebook to preview a sales refunds table, split it into train and predict tables, run a time-series check and train an AutoML forecast, keeping what it decided in Markdown memory files. That is the agent tooling edge; the same toolkit is then started as an MCP server over stdio and over streamable HTTP and reached from a coding agent and from Joule Work's desktop, which is where the live demo partly fails and the failure is shown.published 2026-09-22
From Intent to Arcade: Building Joule*Bert with Joule Studio and AI-Assisted Pro CodeFrom Intent to Arcade: Building Joule*Bert with Joule Studio and AI-Assisted Pro CodeA Devtoberfest 2026 AI-track session that takes a Q*bert-style arcade game, written as an Easter egg for colleagues, through every phase Joule Studio runs: intent analysis with its clarifying questions, a product requirements document, a specification with its task list, generated code, generated tests and deployment. The knowledge-graph lookup reports that the game maps to no enterprise process and the flow adjusts, and the presenter edits the requirements by hand when the first version declares sound out of scope. He then pushes the project to GitHub, clones it and initialises the jl command line so that Joule Studio's MCP servers are wired into the coding agent in VS Code, where mobile arrow keys were added afterwards; applications could not yet be deployed to the Joule Studio runtime, so this one runs on Kyma.SAP Developers, published 2026-10-01
Governance at Scale: Why a Clean Core Is What Makes Agent Governance EnforceableGovernance at Scale: Why a Clean Core Is What Makes Agent Governance EnforceableClosing a six-part clean-core series, the post argues that SAP's two agentic-AI governance mechanisms only work against a documented tool surface: MCP Gateway in SAP Integration Suite auto-translates API specifications into MCP tool schemas augmented with natural-language semantics, carrying forward existing authentication, authorization and rate limiting into the agentic layer, while AI Agent Hub, built on SAP LeanIX, discovers and runs verification assessments on agents, LLMs and MCP servers before the resulting badge is read by Joule Studio or SAP Integration Suite to decide what is trusted. Its argument holds the mechanisms up against an undocumented internal interface, where the auto-generated schema and the verification badge have nothing real underneath them, and against a consolidated CAP service exposing only released APIs, where both do.DEEPA_DORAIRAJ, published 2026-09-21
How to implement business AI capabilities in ABAP CloudHow to implement business AI capabilities in ABAP CloudA Devtoberfest 2026 session (ABA-32) on embedding AI in an ABAP Cloud application and on making SAP Joule aware of what that application can do. The first demo implements two features of a RAP travel app with the ABAP AI SDK: a determination on an uploaded receipt that sends the image together with the allowed categories and transaction types to a model reached through an intelligent scenario, then writes the extracted description, category, currency and amount back to the expense entity; and a recommendation function, triggered by a side effect on the trip title, that proposes a trip type with a score the Fiori UI sorts the proposals by and leaves the business user to accept — which is LLM integration from ABAP code. The second demo builds a capability for SAP Joule in the Joule Studio code editor in VS Code, with a scenario that extracts a trip title from the user's intent and calls an OData service in the ABAP backend through a BTP destination, deploys it and asks the Joule test UI for the trip's details; the session states that agents reach ABAP only over governed remote APIs — OData, SQL and MCP — and that automatic grounding of custom OData services in the knowledge graph is upcoming rather than available.published 2026-10-05
Meet Joule Studio and Build your First AI AgentMeet Joule Studio and Build your First AI AgentA Devtoberfest 2026 session from the product manager of Joule Studio, walking one agent through the whole path the tool defines: an intent, the problem statement and goals a coding agent derives from it, the business context it takes from the SAP knowledge graph, and the solution type it proposes with a stated intent fit. The demo is a collections agent that drafts dunning emails: the product requirements document names the metrics and the human-in-the-loop review, the technical specification lists the artifacts to build — the MCP tools among them — and the build is followed by a business test run against business partner 10001, technical unit tests, and deployment to the managed runtime, where the agent and its MCP server appear as deployed assets. Two points shape how it is used rather than what it produces: every asset is a file, editable in the web editor or in the code view and pushable to git, and the same solution can be built from local pro-code tools; n8n workflows are stated as delivered out of the box inside the studio.SAP Developers, published 2026-09-24
Real Pro-Code Development with Joule StudioReal Pro-Code Development with Joule StudioA Devtoberfest 2026 AI-track session that runs one agent build across both of Joule Studio's paths. Playing a procurement manager, one presenter states the intent in the browser, answers the clarifying questions, watches the fit-gap analysis name the S/4HANA APIs the use case needs, and pushes the intent and requirements document to GitHub; the other clones it and carries on in a terminal, logging in with the jl command line against her landscape URL and running its init command to fetch Joule Studio's skills for OpenCode. From there the coding agent writes the specification, the discovered API specifications, the MCP translation and the agent's Python file with its model configuration and evaluation scenarios, which is the edge to Agent tooling; the skills themselves, which SAP says not to edit but which a developer may add to, are the edge to Agent guidance.SAP Developers, published 2026-10-01
SAP BDC and AI: A Perfect MatchSAP BDC and AI: A Perfect MatchA Devtoberfest 2026 data-track session (DAT-22) in which Business Data Cloud's CTO sets out how SAP Business Data Cloud feeds SAP's business AI platform: data products whose primary persistence is an object store and whose semantics are kept with them, the one domain model that links a supplier in Ariba to a supplier in S/4HANA, and the SAP knowledge graph, a triple store running in SAP HANA Cloud and queried with SPARQL, that SAP Joule consults before answering. The live demo asks Joule for the available data products and the top ten open purchase orders, with the data product query agent compiling SQL over virtual tables and a space rendering a dashboard, which is the cataloguing side of this: a central registry, the Unified Customer Landscape, that every activated data product is registered in. The Q&A covers where the knowledge graph lives and what a customer may extend, base policies that turn into filter conditions on an agent's queries, and Joule Studio as the place to build your own agents and data products.SAP Developers, published 2026-09-29
SAP BDC in Business AI Platform: Contextualize and ReasonSAP BDC in Business AI Platform: Contextualize and ReasonA Devtoberfest 2026 session (DAT-31) that places SAP Business Data Cloud as the contextualise-and-reason pillar of the SAP Business AI Platform, between the build pillar of Joule Studio and Integration Suite and the govern pillar of AI Agent Hub. It covers data products described with open resource discovery and core schema notation metadata, a data product creation agent that assembles a derived data product for overdue billing documents from the billing document and customer products and shows its lineage and fields; the data products a line-of-business application brings being readable by Joule without a BDC subscription; and the SAP-managed knowledge graph that maps business entities and their relationships for agents, held as graph data in the SAP HANA Cloud knowledge graph engine as RDF triples queried with SPARQL. The closing demo has SAP Joule answer head-count and purchase-order questions, with the generated SQL passed through a data product query that adds the restriction policy defined for that user in cloud identity services, which is why the session stands for cataloguing and lineage as much as for the products it names.SAP Developers, published 2026-10-06
SAP Joule: Invoice Agent for Faster Supplier Invoice Release in SAP S/4HANA Accounts PayableSAP Joule: Invoice Agent for Faster Supplier Invoice Release in SAP S/4HANA Accounts PayableWalks through building an Invoice Release Agent in SAP Joule: a custom OData service surfaces supplier invoices whose S/4HANA Flexible Workflow approval is overdue, the agent ranks them by invoice value, aging, payment tenor or a combined score, and a Joule Skill calls the standard SAP AP Invoice OData API to release them, re-prioritizing the remainder after each action. It sets out the detect-prioritize-act loop and the four ranking criteria in enough detail to rebuild the pattern, and leaves authorization, auditability and segregation-of-duty controls to the reader's own design.idc, published 2026-08-11
SAP's API Policy on AI Agents: What It Prohibits, Allows, and Leaves OpenSAP's API Policy on AI Agents: What It Prohibits, Allows, and Leaves OpenReading SAP's API Policy v.4.2026a and its FAQ against the reference architectures, the piece narrows the widely repeated claim that SAP bans AI to what Section 2.2.2 actually restricts: an agent planning and chaining API calls outside an SAP-endorsed route, closed off further by Section 3's block on bypassing API Controls through custom code. It compares the endorsed and allowed paths — the MCP Gateway on SAP Integration Suite reaching S/4HANA and ECC through RFC-based back ends under a 15-tool, 200,000-token limit, SAP-provided MCP servers through Joule Studio, the A2A protocol, customer-operated MCP servers allowed under FAQ Question 56, and the ADT MCP Server, whose 18 tools work only on development artifacts and are documented as not intended for productive use. It leaves open whether Section 2.2.2 covers ABAP code that calls a model from inside the system, since enforcement is defined at the API-traffic level rather than by a list of tools.published 2026-08-31
September 15th: What's New on developers.sap.comSeptember 15th: What's New on developers.sap.comThomas Jung walks through a week of changes aimed at agentic consumption of developers.sap.com itself: the hosted MCP server gains a published .well-known/mcp.json manifest, curated tools (search_tutorials, get_tutorial_step with a Markdown format option, search_channels), and a listing in the ORD catalog beside its OData, REST and GraphQL APIs. It also documents the site's existing A2A endpoint for the first time and ships a way to turn a tutorial into an installable, cryptographically signed AI Skill with a generated verification script, both concrete instances of agent tooling; separately, five broken GraphQL example queries were fixed and the full SDL is now served, and new sitemap.md, llms.txt fixes and ORD entries round out the site's own metadata publishing.published 2026-09-15
What's in Daniel Hutzel's brain?What's in Daniel Hutzel's brain?A Devtoberfest 2026 session in the building-apps track, in which Daniel Hutzel and DJ Adams work through what the CAP team is building for agents. A bookshop gets an MCP endpoint beside its OData one from a single annotation, with the plugin writing the coding agent's configuration itself; in the xt-travels sample five services that know nothing about each other are served over MCP and an agent does the orchestration the hand-built UI and its value helps used to do. Hutzel then annotates a service as an agent so the ReAct loop and the choice of model move into the CAP service, and finally drops MCP for a skill that teaches the agent CQL instead, which is where the edges to Serve your application over MCP, Agent tooling and Agent guidance come from. The mechanism under all of it, a plugin file in the package root and event handlers that can be added to any CAP service, is the edge to Plugin authoring.published 2026-10-02
What's new in SAP Cloud Application Programming Model – September 2026What's new in SAP Cloud Application Programming Model – September 2026The September 2026 release turns CAP services into A2A-callable agents through the new @cap-js/agents plugin, which generates MCP tools from a service's entities and actions, adds an @agent.hitl approval gate and can read an AGENTS.md/SKILL.md pair for custom behaviour. The @cap-js/mcp plugin for Node.js and com.sap.cds:cds-adapter-mcp for Java reach general availability, serving describe, query and call tools over existing CAP authentication and authorisation. It also names capire/skills, SAP's curated skill library that teaches coding agents current CAP practice.published 2026-09-24
🔴 Your Agent Wants to Call SAP. Now What?🔴 Your Agent Wants to Call SAP. Now What?Devtoberfest 2026 session INT-23, on what belongs between an agent and an SAP system. A solution advisor and an enterprise architect point Joule Desktop extensions at MCP servers built on API Management inside SAP Integration Suite, and ask it for business partners, orders and line items from an S/4HANA demo system; asked for the ten largest customers it answers from a sample of 200 orders rather than reading the whole table. They then open the API proxy behind it, where a spike arrest policy caps the backend at ten calls per second and the MCP server is configured to expose only the GET resources of the business-partner API so that an agent cannot create one, which is where the edges to Serve your application over MCP and Rate limiting hold. For totals across all sales orders they send the agent to SAP Business Data Cloud instead of the transactional system, since it aggregates and is read-only, and back through Integration Suite for writes such as journal entries.SAP Developers, published 2026-09-30
🟣 AI Eating Brains! How I Fed Mine to Joule Work Desktop🟣 AI Eating Brains! How I Fed Mine to Joule Work DesktopDevtoberfest 2026 session AI-22. Mathias Kemeter connects his Obsidian note vault to Joule Work Desktop with VaultGate, a local HTTP MCP server he wrote in his spare time because the desktop's sandbox accepts HTTP MCP servers and not stdio ones, and because he wanted the vault reached through Obsidian's own command line rather than by direct file access. The demo builds a Space of management tasks from notes and mailbox together, has the assistant pick up a conventions file in the vault without being told where the reporting rules are, finds a conflict the notes never name by semantic search over local embeddings, and appends a task to a sprint note behind a write confirmation, which is where the edges to Joule Work and Agent workspaces hold. The second half is about the guardrails he had AI choose for the project it then had to live inside: tests and coverage, type safety, static checks, dependency and secret scanning, blocking on commit, on push and in CI.published 2026-10-01
Be the First to Understand Joule Work's Agentic ThinkingBe the First to Understand Joule Work's Agentic ThinkingA Devtoberfest 2026 session on the agent harness behind agentic thinking in Joule Work: generic operating-system tools in a gVisor-isolated sandbox instead of hand-wired domain functions, a react loop, a warm pool of sandboxes torn down after each task, tenant-separated skill loading, and placeholder tokens the agent never sees swapped for real credentials at the egress path. A skill is a folder with a SKILL.md carrying metadata and instructions, and where it reaches a system, an SAP connectivity JSON that declares each connector by ORD ID with its guardrails — GET allowed by default, POST, PUT, PATCH and DELETE requiring human approval, everything else blocked. The demo uploads a cost-centre skill through the skills view, then has agentic thinking delegate over A2A through Agent Gateway to a custom agent built in Joule Studio, whose MCP tools read cost centres from an S/4HANA Cloud system, and ends by turning the result into a PDF; the Joule Studio trace is what shows the tool call actually happened.SAP Developers, published 2026-10-15
Supercharge Joule Desktop: Connect Any SAP API in Minutes with MCP GatewaySupercharge Joule Desktop: Connect Any SAP API in Minutes with MCP GatewayDevtoberfest 2026 session INT-13, in which an SAP solution advisor connects a live S/4HANA system to the desktop experience of Joule Work through the MCP gateway in SAP Integration Suite. One OpenAPI specification covering business partners, sales orders, material stock, billing and purchase requisitions is uploaded as an MCP server artifact, deployed to integration cell with XSUAA inbound authentication, and registered as an MCP connector; the destination behind it is then switched from basic authentication to principal propagation, and the Cloud Connector monitor shows the same question arriving as the presenter's own user. That demonstration is the serve your application over MCP edge, and the second half is Joule Work itself: a Markdown skill that fixes the steps and format of a customer briefing, and a Space the skill rebuilds with each run.SAP Developers, published 2026-09-23
@sap-ai-sdk/orchestration@sap-ai-sdk/orchestrationThe Orchestration Service client of SAP's AI SDK, which applies templating, grounding, filtering and masking around a model call rather than in the application.sap official, cap
Orchestration Service client for templating, grounding and masking
SAP BTP AI Best Practice Demo - Content Filtering - CAPSAP BTP AI Best Practice Demo - Content Filtering - CAPA CAP backend-sample and fullstack-sample that call SAP AI Core with content filtering switched on and off, to compare the filtered and unfiltered response.sap official, cap
CAP samples comparing SAP AI Core responses with content filtering
Every CDS service in both samples is marked `@requires: 'any'` and left publicly accessible, which the README says simplifies the testing and deployment process, and it asks for proper authentication mechanisms to be applied in their place.
sap-btp-ai-best-practicessap-btp-ai-best-practicesSAP's collection of sample implementations for AI-powered applications on SAP BTP, one directory per topic under best-practices, in TypeScript, Python, Java and CAP.sap official, btp
Sample implementations for AI-powered applications on SAP BTP
@sap-ai-sdk/orchestration@sap-ai-sdk/orchestrationThe Orchestration Service client of SAP's AI SDK, which applies templating, grounding, filtering and masking around a model call rather than in the application.sap official, cap
Orchestration Service client for templating, grounding and masking
SAP BTP AI Best Practice Demo - Data Masking - CAPSAP BTP AI Best Practice Demo - Data Masking - CAPA CAP backend-sample and fullstack-sample that generate an email through SAP AI Core with an anonymize flag on the request; the README's example prompt carries an email address.sap official, cap
CAP samples generating email through SAP AI Core with anonymization
Every CDS service in both samples is marked `@requires: 'any'` and left publicly accessible, which the README says simplifies the testing and deployment process, and it asks for proper authentication mechanisms to be applied in their place.
sap-btp-ai-best-practicessap-btp-ai-best-practicesSAP's collection of sample implementations for AI-powered applications on SAP BTP, one directory per topic under best-practices, in TypeScript, Python, Java and CAP.sap official, btp
Sample implementations for AI-powered applications on SAP BTP
Connecting SAP Integration Suite with SAP Document AI and SAP Business AIConnecting SAP Integration Suite with SAP Document AI and SAP Business AIWalks through a Cloud Integration flow in SAP Integration Suite that pulls PDF purchase orders from a Gmail mailbox, extracts header and line-item fields through SAP Document AI's job-submission and polling API, and validates the result with SAP AI Core's generative AI hub using GPT-5-mini as a quality check rather than the source of the extracted values. Groovy steps combine the Document AI confidence score and the model's score into one figure that routes low-scoring purchase orders to manual review, with no S/4HANA posting performed.subhojitSaha, published 2026-08-24
How to Use n8n for Agentic Integrations and Cool InsightsHow to Use n8n for Agentic Integrations and Cool InsightsA Devtoberfest 2026 session on one use case built with n8n, the workflow automation tool SAP announced a partnership with in May 2026: judging the attachments of an automotive warranty claim. A Fiori upload table carries named evidence rows — this line is the invoice, this one the quote — and each uploaded file goes as base64 to an n8n webhook together with the claim context: the position and diagnosis codes, the parts, the amounts and the conversation with the dealer. The flow splits PDF from image, tries text extraction before falling back to a vision prompt, and calls Anthropic's Sonnet with a prompt that names the role, the input, the scoring rules and the response shape; the demo shows a matching invoice scoring 95%, an unrelated dent-removal invoice scoring 5% with the mismatches listed, and a photograph of a rusted suspension arm accepted while an engine bay is rejected against the same claim. The second half is the n8n editor itself — nodes, credentials, the test trace with each node's input and output — and the closing advice is human-in-the-loop flows at the edges of a process rather than automatic adjudication inside it.SAP Developers, published 2026-09-25
What a Second Document AI Workshop Taught Me About GovernanceWhat a Second Document AI Workshop Taught Me About GovernanceFrom questions raised at a second SAP Document AI workshop: the retention period for extracted data is a setting in the service's administration rather than a fixed property, with the source file deleted after processing and the decommissioning period for the extracted data configured to the customer's own policy. It also draws a boundary readers kept merging: Document AI extracts across languages but does not translate, so turning a document into a reviewer's own language is a separate step, pointing on BTP at SAP Translation Hub.Nicolas_BacaStorni, published 2026-08-25
Why Our First Document AI Exercise Was a Résumé, Not an InvoiceWhy Our First Document AI Exercise Was a Résumé, Not an InvoiceWorking through a SAP Document AI workshop, the author explains why its exercises teach extraction on a résumé before an invoice: a résumé has no stable layout to map fields against, so it forces the point that the tool extracts from a written description of a schema rather than from position-matched templates. Describing the schema is shown to carry real judgment — what a missing end date should become, what a vague field name like "number" costs the model, who in a company should decide these things — and the piece argues that preparing for a project is chiefly a business task of writing down knowledge that otherwise stays in people's heads.Mariano_BacaStorni, published 2026-09-03
UI5con_2026_MCPAppsUI5con_2026_MCPAppsFive runnable MCP servers whose tools return a prebuilt UI beside their text result, from plain HTML through UI5 Web Components and UI Integration Cards to a UI5 linter workflow and an ARC-1 transport review.independent, ui5
Five MCP servers whose tools return a prebuilt UI
Neither the package manifest nor the repository tree declares a licence.
@cap-js/ai@cap-js/aiOut-of-the-box value-help recommendations powered by SAP RPT-1, SAP AI Core as a standard CAP service, and an experimental local embedding model behind SQLite's vector functions.sap official, cap
Value-help recommendations and SAP AI Core as a CAP service
@jerome-benoit/sap-ai-provider@jerome-benoit/sap-ai-providerA provider for the Vercel AI SDK that speaks to SAP AI Core through its orchestration or foundation models API, so an application written against that SDK runs on SAP's models without being rewritten.independent, cap
Vercel AI SDK provider for SAP AI Core
@sap-ai-sdk/ai-api@sap-ai-sdk/ai-apiThe AI Core API surface of SAP's AI SDK, covering deployments, configurations and the rest of the control plane, as opposed to calling a model.sap official, cap
AI Core control plane API surface of SAP's AI SDK
@sap-ai-sdk/core@sap-ai-sdk/coreThe shared internals that the packages of SAP's AI SDK build on, which its own readme says are not meant to be depended on directly.sap official, cap
Shared internals of SAP's AI SDK
@sap-ai-sdk/foundation-models@sap-ai-sdk/foundation-modelsThe package of SAP's AI SDK that calls foundation models through SAP AI Core and the Generative AI Hub, and the one an application actually reaches for.sap official, cap
Client for foundation models via SAP AI Core and Generative AI Hub
@sap-ai-sdk/orchestration@sap-ai-sdk/orchestrationThe Orchestration Service client of SAP's AI SDK, which applies templating, grounding, filtering and masking around a model call rather than in the application.sap official, cap
Orchestration Service client for templating, grounding and masking
@sap-ai-sdk/prompt-registry@sap-ai-sdk/prompt-registryA package for keeping prompts in SAP's prompt registry rather than in the codebase, so changing a prompt is not a deployment.sap official, cap
Client for prompts kept in SAP's prompt registry
@saptarishi/cds-plugin-llm@saptarishi/cds-plugin-llmA CAP service kind that puts eleven model providers behind one chat, stream, embed and batch interface, so which backend a service calls is a profile in package.json rather than a change to the handler.independent, cap
CAP service kind putting eleven model providers behind one API
@shahid.la/cds-db-nlquery-mcp@shahid.la/cds-db-nlquery-mcpMCP server that answers plain-English questions against the entities in db/schema.cds — it discovers the schema, generates CQN-backed SQL with real joins, and returns rows.independent, cap
MCP server answering plain-English questions against a CDS schema
@ui5/webcomponents@ui5/webcomponentsSAP's UI elements as standard custom elements, ui5-button among them, implementing SAP Fiori design and usable from plain HTML, React, Angular or Vue.sap official, ui5
SAP's UI elements as standard custom elements
Not where the UI library itself is expected to bind data: the documentation leaves binding a component's content and events to the host framework. Not from a CDN script tag: the project publishes none, so it assumes a build step.
abap-agent-runtimeabap-agent-runtimeAn ABAP-native MCP endpoint inside the SAP system: an agent reads the screen of an opted-in abap2UI5 app as structured data, fills its fields and fires its events as the logged-on SAP user.community org, abap
ABAP-native MCP endpoint that drives opted-in abap2UI5 apps
Stateful apps cannot be operated: a stateful session lives in one HTTP request and an MCP call is one request, so `app_start` refuses them. The repository carries no release yet and states the addon has not been run in production use, the 7.02 downport is lint-only with no 7.02 system having run it, and the ABAP Cloud HTTP service setup is written down but still to verify on your own system.
cap-java/cds-aicap-java/cds-aiBridges CAP Java to SAP AI Core and adds AI-powered field recommendations to Fiori draft UIs, as two alpha Maven modules with mock fallbacks for local development.sap official, cap
Bridge from CAP Java to SAP AI Core, with Fiori field recommendations
ui5-cap-ai-featuresui5-cap-ai-featuresBoth halves of an AI text feature in one package: a CDS plugin serving an AIService over SAP AI Core's Orchestration Service, and the be.wl.ai UI5 control library that calls exactly that service.independent, cap
CDS plugin and UI5 controls for AI text features via SAP AI Core
It is text processing, not an agent: there is no tool calling, no retrieval and no conversation. It needs SAP AI Core, and the backend half needs [@sap-ai-sdk/orchestration](artifact:sap-ai-sdk-orchestration) installed alongside it — an optional peer, so a frontend-only install stays lean.
ai-enabled-integrations-codejamai-enabled-integrations-codejamA CodeJam that processes customer support requests in SAP Cloud Integration with an LLM in SAP AI Core, then exposes APIs to an agent as MCP servers through API Management's MCP Gateway.sap official, btp
CodeJam on LLM-backed integration flows and MCP via API Management
aicore-genai-samplesaicore-genai-samplesA collection of sample applications that show how to use SAP AI Core's Generative AI Hub capabilities, each with its own README and meant for enablement, not production use.sap official, btp
Sample apps for SAP AI Core's Generative AI Hub capabilities
cap-agent-ui5-webui-samplecap-agent-ui5-webui-sampleA worked example of the cap-agent-ui5-webui plugin, shown running locally, in hybrid mode against SAP AI Core, and deployed.independent, cap
Worked example of the cap-agent-ui5-webui plugin
Neither the package manifest nor the repository tree declares a licence.
Sample CAP Application with SAP Cloud SDK for AISample CAP Application with SAP Cloud SDK for AIA CAP backend-sample and fullstack-sample pair that call SAP AI Core through the SAP Cloud SDK for AI: a capital-of-country question and a templated chat completion.sap official, cap
CAP samples calling SAP AI Core through the SAP Cloud SDK for AI
Every CDS service in both samples is marked `@requires: 'any'` and left publicly accessible, which the README says simplifies the deployment process, and it asks for proper authentication mechanisms to be applied in their place.
sap-btp-ai-best-practicessap-btp-ai-best-practicesSAP's collection of sample implementations for AI-powered applications on SAP BTP, one directory per topic under best-practices, in TypeScript, Python, Java and CAP.sap official, btp
Sample implementations for AI-powered applications on SAP BTP
SAP-samples/btp-cap-genai-ragSAP-samples/btp-cap-genai-ragA CAP reference application on SAP BTP for GenAI email support, built on SAP AI Core's Generative AI Hub and retrieval-augmented generation over SAP HANA Cloud's Vector Engine.sap official, cap
CAP reference app for GenAI email support with RAG on HANA Cloud
SAP-samples/codejam-cap-llmSAP-samples/codejam-cap-llmA CodeJam that builds a CAP-based job posting service exposing SAP AI Core's generative AI capabilities, including vector embeddings, through an OData API.sap official, cap
CodeJam on a CAP job posting service using SAP AI Core over OData
SAP-samples/codejam-code-based-agentsSAP-samples/codejam-code-based-agentsA CodeJam teaching code-based AI agents in Python or TypeScript, from a first agent through custom tools to deployment on SAP BTP and integration with Joule.sap official, btp
CodeJam on code-based AI agents in Python or TypeScript on SAP BTP
teched2025-DA266teched2025-DA266SAP TechEd 2025 session material for two SAP Databricks scenarios in SAP Business Data Cloud: cashflow forecasting and payment-delay prediction with SHAP explanations.sap official, btp
TechEd 2025 material on two SAP Databricks scenarios in BDC
ui5-ai-features-demoui5-ai-features-demoA product-review screen on CAP and SAPUI5 that consumes ui5-cap-ai-features and contains no library source at all — written to prove the package is enough on its own.independent, cap
Product-review screen consuming ui5-cap-ai-features on CAP
Neither the package manifest nor the repository tree declares a licence.
🟣 Bring AI into Your ABAP Applications with ISLM: Build, Integrate, and Consume AI Natively🟣 Bring AI into Your ABAP Applications with ISLM: Build, Integrate, and Consume AI NativelyDevtoberfest 2026 session AI-11, from the product manager and product owner of Intelligent Scenario Lifecycle Management: ISLM as the place an AI use case is registered, deployed and operated inside an SAP S/4HANA system, and what has been added to it for generative AI. The demo builds a custom scenario end to end — an intelligent scenario of type generative AI, a connection to SAP AI Core, a GPT-5 mini model, an orchestration flow template that configures data masking, input and output filtering and translation into the user's logged-in language, then prompts with a grounding query over five companies' return-policy documents — and consumes it from an ABAP report through the ABAP AI SDK, which calls the orchestration API and returns the summary as JSON. Two newer pieces follow: SAP RPT, the relational pre-trained transformer for tabular prediction, called with context rows and query rows and answering with a confidence value and no training step; and system-triggered agents, where an application job in the ABAP system starts a billing-creation agent that reaches S/4HANA OData APIs as tools through the agent gateway with principal propagation. The closing availability matrix states the generative AI capabilities for on-premise and private cloud from the 2021 release, with SAP notes to implement on lower ones.SAP Developers, published 2026-09-24
🟣 Enable Your AI application to Run on Multiple Models to Ensure Resilience and Compliance🟣 Enable Your AI application to Run on Multiple Models to Ensure Resilience and ComplianceDevtoberfest 2026 session AI-21. A cross product owner on the SAP AI Core team gives the reasons an AI application ends up on more than one model — a model down at the hyperscaler, a data centre that is not allowed to run it, a version being deprecated — and runs a joke app that answers from Gemini 2.5 Flash in EU10 and from Claude 4.5 Haiku in Asia-Pacific. The walk through AI Core's parts is where the edges hold: prompts kept as templates in the prompt registry, imported through the launchpad and the APIs or synced as YAML from a git repository (Prompt management); the evaluation service scoring one prompt across up to four models per run on built-in or model-as-judge metrics, and the prompt optimizer generating variants against one metric (Agent evals); and the orchestration service harmonising the providers behind one message format with a priority-sorted fallback list (LLM integration). The session also states where a metric has to be designed rather than picked: an exact-JSON match scores an email generator at zero, so the joke app's custom metric judges topic and family-appropriateness instead.SAP Developers, published 2026-10-01
Bring "developers.sap.com" Into Your Dev Environment: SAP Developer Context for AI Coding AgentsBring "developers.sap.com" Into Your Dev Environment: SAP Developer Context for AI Coding AgentsDevtoberfest 2026 session FA-11: a tour of the rebuilt developers.sap.com by the team that runs it, from the administration screens behind the home page to the agent interfaces in front of it. The site is a CAP application on SAP HANA Cloud and SAP BTP, open-sourced in full, and the half of the session that places it here is how its content reaches an agent: every tutorial is also served as markdown, a coding agent is detected and given that version instead of the HTML, a well-known agent card advertises what the site offers, and SAP Developers MCP answers questions about tutorials, news and a reader's own completion history in Claude Code and Joule Desktop, over OAuth or a personal access token. Behind the answers sit two things working together: a knowledge graph in the HANA database over 1400 tutorials and 11,000 concepts, and SAP AI Core for the Joule interactions and the embeddings, with the agent harness written for the site rather than taken off a shelf. Where neither MCP nor A2A is available, the same content is published as skills.SAP Developers, published 2026-09-25
Build Your AI Agent prototype for free using SAP CAP, LangChain and Ollama cloudBuild Your AI Agent prototype for free using SAP CAP, LangChain and Ollama cloudA CAP Node.js service action calls an Ollama-hosted model through LangChain's ChatOllama, with the API key read from a .env file instead of any destination or credential binding — the hand-rolled-key route SAP's own SDK call exists to avoid. The same handler then builds a LangChain agent with createAgent and one mock tool, so a question about restaurants triggers a tool call and any other question does not, independent of the @cap-js/agents plugin's A2A-served harness.rajeevgoswami1, published 2026-08-29
Building a Pro-Code Agent on SAP BTP — Part 2: The Agent (LangGraph + GenAI Hub)Building a Pro-Code Agent on SAP BTP — Part 2: The Agent (LangGraph + GenAI Hub)Part 2 of a three-part series builds a separate, non-CAP Node/TypeScript agent that calls the Part 1 CAP service's MCP tools and reasons through SAP AI Core via @sap-ai-sdk/orchestration, using LangGraph's createAgent ReAct loop for provider-agnostic model calls across GPT, Claude and Gemini deployments. It explains why the tool's Zod schema and model-facing description are the agent's actual contract, the gotcha between a foundation-models deployment and an orchestration-scenario one, and how edge-case testing replaced date inference with a server-computed daysOverdue field after the agent wrongly called a delayed order 'on schedule'.rodfior, published 2026-08-20
Building Agentic Workflows with n8n in Joule StudioBuilding Agentic Workflows with n8n in Joule StudioA Devtoberfest 2026 session, JS-31, on SAP-managed n8n inside Joule Studio, shown as a labs preview on a blocked-invoice scenario: a solution made of assets, an intent refined by the web agent into requirements and then a specification, and the generated workflow opened in the n8n editor that Joule Studio connects to its own n8n tenant, with the generated files committed through GitHub. The nodes carry the other edges — the human-in-the-loop node is built on SAP Task Center, and SAP AI Core is where model providers are configured centrally instead of per developer. It ends in Joule chat, where the approval task the workflow created is listed, opened and completed.SAP Developers, published 2026-10-08
Building Code-Based AI Agents on the BAIPBuilding Code-Based AI Agents on the BAIPA Devtoberfest 2026 session on writing the agent yourself rather than configuring one, which is the agent tooling edge: an agent is a model with tools, skills, memory and context around it, and the session reads the code of one built in CAP as a LangGraph state graph, with an agent node that calls the model, a tools node holding the MCP clients, and the edge between them looping until no tool is needed. It then shows the shorter route, a CDS service annotated for the agent protocol with its instructions and skills in an agent.md file. SAP AI Core carries both, with the CAP service in front of it as middleware so the credentials never reach the browser, and that is what cap-agent-ui5-webui is demonstrated against: added as a dependency, it gives the annotated service a UI5 chat interface that can be deployed rather than only previewed. A Python multi-agent application on Cloud Foundry, where an orchestrator delegates to per-stack specialists so no single model sees every tool, closes the session.published 2026-10-09
CAP in the age of AI #3 - from MCP to A2ACAP in the age of AI #3 - from MCP to A2AThe episode that finishes @cap-js/mcp with authorisation and then crosses to agents. @requires and @restrict are the same annotations CAP already has, and the plugin reads them to filter the tool list and the describe output per user and per request, so a model is not told about entities the caller cannot reach — where the OData metadata of the same service lists every exposed entity regardless. The second half annotates a service with @agent from @cap-js/agents, which serves it over A2A with an agent card and a chat preview, runs first against a mocked response and then against SAP AI Core bound with cds bind and a hybrid profile, and reuses the MCP adapter's tools rather than defining new ones.published 2026-08-21
CAP in the age of AI #5 - agents wrap-upCAP in the age of AI #5 - agents wrap-upThe agents topic closed out on what sits around the agent rather than inside it, now that the documentation has moved into capire. The chat preview's session is persisted in the database rather than kept in memory, traces go out through @cap-js/telemetry to a self-hosted MLflow under the OpenTelemetry conventions for generative AI, and the panel is explicit that @agent reuses the MCP adapter's code programmatically without serving an MCP endpoint. It ends in the xtravels sample, where a travel agent delegates to a hotel-service sub-agent and reaches a separately started flight service over MCP through the service binding CAP resolves anyway. Earlier in the hour the same agent answers from a model running on the presenter's laptop, reached as an OpenAI-compatible endpoint.published 2026-10-02
CAP in the age of AI #6 - vector embeddingsCAP in the age of AI #6 - vector embeddingsVector embeddings put to work in the incident management sample, answering whether this problem has been seen before. An embedding element of type Vector is calculated and stored from the vector_embedding function, so the database refreshes it whenever the text behind it changes, and cosine_similarity compares it with a freshly embedded string; because a title says little, a scheduled action first summarises each incident's conversation through a model on SAP AI Core and the embedding is taken from that summary. The similar-incidents list is a view joining incidents to themselves, with order and limit set in a before handler on the expanded query, and the vector element excluded from the service, since OData knows nothing about vectors — on SAP HANA Cloud the function is native, while SQLite needs packages added to the project.published 2026-10-09
CAP in the age of AI #7 - vector embeddings wrap-upCAP in the age of AI #7 - vector embeddings wrap-upThe same embeddings model taken from SQLite to SAP HANA Cloud, then handed to an agent. Locally it is @cap-js/ai that adds the vector functions SQLite lacks and downloads an all-MiniLM L6v2 ONNX model on first start; on HANA natural language processing has to be switched on, the model name is a mandatory argument, the vectors it returns are 768 dimensions, and dimensions must match on both sides of a comparison. HANA itself can be given SAP AI Core as a remote source for other embedding models, which a CDS.ql expression has to inject because the postfix notation only takes what the CDS model holds, and the hour ends with a search function over cosine similarity exposed over @cap-js/mcp, so an agent asked about low output from solar panels calls semantic search instead of writing the query itself.published 2026-10-16
CAP: The Road Ahead - Simplicity in an AI EraCAP: The Road Ahead - Simplicity in an AI EraDevtoberfest 2026 session CAP10, from the head of SAP's Cloud Application Programming Model: forty minutes of live building rather than slides, on the argument that the cheapest code in an AI project is code nobody has to generate. Claude Code builds an invoice-matching workspace from cds init onwards, and each step is one the framework answers rather than the agent: a 42-line domain model that @sap/cds serves as OData and Fiori elements with a status flow, data privacy and change-tracking annotations that produce audit log entries in the console and would reach the SAP Audit Log service once deployed, an imported S/4HANA supplier data product, and semantic matching from a local embedding model after @cap-js/ai is added. The last three steps are annotations: @mcp beside @odata turns the service into an MCP server that a second Claude session queries through @cap-js/mcp, an agent annotation adds the A2A protocol with an agent card and an audit log entry for the agent task, and a skill file gives the agent the matching policy.SAP Developers, published 2026-09-25
Connecting SAP Integration Suite with SAP Document AI and SAP Business AIConnecting SAP Integration Suite with SAP Document AI and SAP Business AIWalks through a Cloud Integration flow in SAP Integration Suite that pulls PDF purchase orders from a Gmail mailbox, extracts header and line-item fields through SAP Document AI's job-submission and polling API, and validates the result with SAP AI Core's generative AI hub using GPT-5-mini as a quality check rather than the source of the extracted values. Groovy steps combine the Document AI confidence score and the model's score into one figure that routes low-scoring purchase orders to manual review, with no S/4HANA posting performed.subhojitSaha, published 2026-08-24
How to implement business AI capabilities in ABAP CloudHow to implement business AI capabilities in ABAP CloudA Devtoberfest 2026 session (ABA-32) on embedding AI in an ABAP Cloud application and on making SAP Joule aware of what that application can do. The first demo implements two features of a RAP travel app with the ABAP AI SDK: a determination on an uploaded receipt that sends the image together with the allowed categories and transaction types to a model reached through an intelligent scenario, then writes the extracted description, category, currency and amount back to the expense entity; and a recommendation function, triggered by a side effect on the trip title, that proposes a trip type with a score the Fiori UI sorts the proposals by and leaves the business user to accept — which is LLM integration from ABAP code. The second demo builds a capability for SAP Joule in the Joule Studio code editor in VS Code, with a scenario that extracts a trip title from the user's intent and calls an OData service in the ABAP backend through a BTP destination, deploys it and asks the Joule test UI for the trip's details; the session states that agents reach ABAP only over governed remote APIs — OData, SQL and MCP — and that automatic grounding of custom OData services in the knowledge graph is upcoming rather than available.published 2026-10-05
UI5 AI Features library is now an npm PackageUI5 AI Features library is now an npm PackageThe release note for a demo turning into a dependency, and it is explicit about why that matters: the earlier version worked, but using it meant cloning the repo and copying both halves, which stops working the moment one of them changes. The post walks through what shipping the CAP plugin and the be.wl.ai control library in one npm version buys — a client and server contract that cannot drift — then lists the eleven controls by group, shows the two-line configuration that activates each half, and notes that this version is actually deployed on BTP behind both a standalone approuter and Work Zone. It closes on the demo repository, written with no library source in it as the test that the package stands alone.published 2026-08-31
UI5 AI Features library is now an npm Package
cds 8, line ended. Affects Runnable code.
Unlocking Tabular Foundation Models: From Experimentation to ProductionUnlocking Tabular Foundation Models: From Experimentation to ProductionA Devtoberfest 2026 session (AI-41) on taking tabular prediction from a playground to a deployment. It sets tabular foundation models against a model trained per use case and against language models reading rows as text, explains in-context learning as examples plus the rows whose target column is missing, and tries SAP RPT 1.6 on a train-delay dataset in the RPT playground, including the column and row importance it returns with the prediction. The same prediction is then rebuilt on SAP AI Core, where the tabular orchestration service plays the part the orchestration service plays for language models: a data destination, a table artifact with its CSN schema and a scenario configuration registered once at design time, after which a predict call passes only the query rows, the scenario configuration and a context-selection strategy. The last demo runs those same calls in natural language through the aicore-skills plugin in a coding agent, and one of them only resolves once the agent is told which scenario configuration to use.SAP Developers, published 2026-10-15
What an AI infra engineer sees when opening SAP's AI stack (1/3): the model gatewayWhat an AI infra engineer sees when opening SAP's AI stack (1/3): the model gatewayA first look at SAP AI Core's Generative AI Hub from someone who builds gateways outside SAP for a living: the orchestration service harmonizes provider calls onto one API (a model name string switches providers, and SAP Note 3437766 carries the current model list, token rates and deprecation dates), runs a fixed, non-reorderable module pipeline — templating, masking through SAP Data Privacy Integration, grounding, content filtering, then the model call — and retries down an ordered fallback list whose entries can each carry their own model and rules. It places @sap-ai-sdk/orchestration among the official JavaScript, Java and Python SDKs, and notes SAP's own LiteLLM provider as a way to put a different gateway in front of it.amishkush, published 2026-09-30
analytics-cloud-rpt15-prediction-demoanalytics-cloud-rpt15-prediction-demoA FastAPI backend that triggers SAP-RPT-1.5 payment-delay predictions from an SAP Analytics Cloud Multi Action and writes the results back to HANA.sap official, btp
FastAPI backend for SAP-RPT-1.5 predictions from SAP Analytics Cloud
hana-apl-apis-runtimeshana-apl-apis-runtimesSQL code examples for every SAP HANA Automated Predictive Library API, plus reference runtimes for scoring an exported model in C++, Java or JavaScript.sap official, btp
SQL examples for the HANA APL APIs plus scoring runtimes
hana-ml-sampleshana-ml-samplesSAP-authored code examples for predictive and machine learning scenarios on SAP HANA, from Predictive Analysis Library SQL to hana-ml Python notebooks.sap official, btp
Code examples for predictive and ML scenarios on SAP HANA
sap-rpt-samplessap-rpt-samplesBruno API collections and code samples for calling SAP's Relational Pretrained Transformer models on SAP AI Core for predictions on tabular data, one subdirectory per model.sap official, btp
Bruno collections and code for SAP-RPT predictions on tabular data
teched2025-DA266teched2025-DA266SAP TechEd 2025 session material for two SAP Databricks scenarios in SAP Business Data Cloud: cashflow forecasting and payment-delay prediction with SHAP explanations.sap official, btp
TechEd 2025 material on two SAP Databricks scenarios in BDC
How We Make Tabular Predictions in 2026 with TabPFNHow We Make Tabular Predictions in 2026 with TabPFNA Devtoberfest 2026 session, AI-32, on why a table needs a model of its own: a language model reasons its way to a plausible number, while a tabular foundation model takes the table as context in one forward pass and learns from its features statistically. That is predictive models from both ends — a scikit-learn-shaped fit and predict in a notebook over a heart-disease data set, predicted class probabilities and SHAP values for which features drove the outcome, then the same model reached as an MCP tool by a coding agent that is asked which current customers are most likely to churn. The presenter is from the lab that publishes TabPFN; the Q&A puts it beside SAP's own tabular foundation model, which she says she has not used.SAP Developers, published 2026-10-08
SAP-RPT-1.5 Is Now Generally Available: More Accurate Predictions, Better Explainability, What's NewSAP-RPT-1.5 Is Now Generally Available: More Accurate Predictions, Better Explainability, What's NewAnnounces general availability of SAP-RPT-1.5, the next release of SAP's foundation model for predictive models over tabular enterprise data: in-context classification and regression with no training step, now with stronger accuracy on high-cardinality columns, a maximum API batch of 512 rows (up from 128), and an "explain by example" mode that surfaces the historical rows that most influenced a prediction rather than ranking feature importance alone. Two variants ship, a default model and a -large model recommended past 100 columns or where accuracy is the priority, and the post previews a coming "tabular orchestration" workflow, targeted for general availability in August 2026, that is meant to select the relevant context rows from millions of historical records automatically.RanjeetKr, published 2026-08-14
Unlocking Tabular Foundation Models: From Experimentation to ProductionUnlocking Tabular Foundation Models: From Experimentation to ProductionA Devtoberfest 2026 session (AI-41) on taking tabular prediction from a playground to a deployment. It sets tabular foundation models against a model trained per use case and against language models reading rows as text, explains in-context learning as examples plus the rows whose target column is missing, and tries SAP RPT 1.6 on a train-delay dataset in the RPT playground, including the column and row importance it returns with the prediction. The same prediction is then rebuilt on SAP AI Core, where the tabular orchestration service plays the part the orchestration service plays for language models: a data destination, a table artifact with its CSN schema and a scenario configuration registered once at design time, after which a predict call passes only the query rows, the scenario configuration and a context-selection strategy. The last demo runs those same calls in natural language through the aicore-skills plugin in a coding agent, and one of them only resolves once the agent is told which scenario configuration to use.SAP Developers, published 2026-10-15
@sap-ai-sdk/orchestration@sap-ai-sdk/orchestrationThe Orchestration Service client of SAP's AI SDK, which applies templating, grounding, filtering and masking around a model call rather than in the application.sap official, cap
Orchestration Service client for templating, grounding and masking
@sap-ai-sdk/prompt-registry@sap-ai-sdk/prompt-registryA package for keeping prompts in SAP's prompt registry rather than in the codebase, so changing a prompt is not a deployment.sap official, cap
Client for prompts kept in SAP's prompt registry
@saptarishi/cds-plugin-llm@saptarishi/cds-plugin-llmA CAP service kind that puts eleven model providers behind one chat, stream, embed and batch interface, so which backend a service calls is a profile in package.json rather than a change to the handler.independent, cap
CAP service kind putting eleven model providers behind one API
SAP BTP AI Best Practice Demo - Prompt Registry - CAPSAP BTP AI Best Practice Demo - Prompt Registry - CAPA CAP backend-sample and fullstack-sample that create, use and delete a template in SAP AI Core's Prompt Registry instead of embedding the prompt in code.sap official, cap
CAP samples using SAP AI Core's Prompt Registry for templates
Every CDS service in both samples is marked `@requires: 'any'` and left publicly accessible, which the README says simplifies the testing and deployment process, and it asks for proper authentication mechanisms to be applied in their place.
SAP BTP AI Best Practice Demo - Prompt Templating - CAPSAP BTP AI Best Practice Demo - Prompt Templating - CAPA CAP backend-sample and fullstack-sample that fill a country placeholder in a prompt template through the SAP Cloud SDK for AI and send it to SAP AI Core.sap official, cap
CAP samples filling a prompt template sent to SAP AI Core
Every CDS service in both samples is marked `@requires: 'any'` and left publicly accessible, which the README says simplifies the testing and deployment process, and it asks for proper authentication mechanisms to be applied in their place.
sap-btp-ai-best-practicessap-btp-ai-best-practicesSAP's collection of sample implementations for AI-powered applications on SAP BTP, one directory per topic under best-practices, in TypeScript, Python, Java and CAP.sap official, btp
Sample implementations for AI-powered applications on SAP BTP
🟣 Enable Your AI application to Run on Multiple Models to Ensure Resilience and Compliance🟣 Enable Your AI application to Run on Multiple Models to Ensure Resilience and ComplianceDevtoberfest 2026 session AI-21. A cross product owner on the SAP AI Core team gives the reasons an AI application ends up on more than one model — a model down at the hyperscaler, a data centre that is not allowed to run it, a version being deprecated — and runs a joke app that answers from Gemini 2.5 Flash in EU10 and from Claude 4.5 Haiku in Asia-Pacific. The walk through AI Core's parts is where the edges hold: prompts kept as templates in the prompt registry, imported through the launchpad and the APIs or synced as YAML from a git repository (Prompt management); the evaluation service scoring one prompt across up to four models per run on built-in or model-as-judge metrics, and the prompt optimizer generating variants against one metric (Agent evals); and the orchestration service harmonising the providers behind one message format with a priority-sorted fallback list (LLM integration). The session also states where a metric has to be designed rather than picked: an exact-JSON match scores an email generator at zero, so the joke app's custom metric judges topic and family-appropriateness instead.SAP Developers, published 2026-10-01
@cap-js/ai@cap-js/aiOut-of-the-box value-help recommendations powered by SAP RPT-1, SAP AI Core as a standard CAP service, and an experimental local embedding model behind SQLite's vector functions.sap official, cap
Value-help recommendations and SAP AI Core as a CAP service
@mi8y/cap-agents-aicore-vectorstore@mi8y/cap-agents-aicore-vectorstoreA LangChain vector store backed by SAP AI Core Document Grounding: the collection holds the documents and makes the embeddings, and the CAP application stores no vectors of its own.independent, cap
LangChain vector store on SAP AI Core Document Grounding
@mi8y/cap-agents-cds-vectorstore@mi8y/cap-agents-cds-vectorstoreA LangChain vector store whose documents and embeddings are CDS entities, so retrieval runs as a CDS query against the database the CAP service is already bound to.independent, cap
LangChain vector store with CDS entities for embeddings
@sap/cds@sap/cdsThe CAP Node.js runtime, which compiles the CDS model, serves the services declared in it, answers generic requests before any handler runs, and loads every cds-plugin.sap official, cap
CAP Node.js runtime
@saptarishi/cds-plugin-llm@saptarishi/cds-plugin-llmA CAP service kind that puts eleven model providers behind one chat, stream, embed and batch interface, so which backend a service calls is a profile in package.json rather than a change to the handler.independent, cap
CAP service kind putting eleven model providers behind one API
@saptarishi/cds-plugin-vector-hana@saptarishi/cds-plugin-vector-hanaVector storage and similarity search on SAP HANA from CAP, packaged as a plugin so the embedding column and the query are modelled rather than hand-written.independent, cap
Vector storage and similarity search on SAP HANA from CAP
sap-knowledge-pipelinesap-knowledge-pipelineA Python library that reads selected records out of an SAP OData service or an SAP HANA SELECT and renders them as chunked documents, each carrying the entity set and business key it came from.independent, btp
Python library turning OData or HANA records into chunked documents
An early alpha release whose README says to pin the version and test recipes against non-production data first. HANA synchronisation is snapshot-only and does not reconcile deleted rows, so a document for a row deleted in the source stays in the index. A recipe's field allow-list does not decide what the service account may read: the README states that field allow-listing does not replace SAP authorization, and that the package infers no SAP module semantics, authorization scope or safe fields from table names. Only Business Partner has a built-in recipe, every other dataset needs a `KnowledgeRecipe` written for it, and [Qdrant](external:qdrant) local mode is stated to be for development and smaller indexes.
SAP BTP AI Best Practice Demo - Vector Embedding (RAG) - CAPSAP BTP AI Best Practice Demo - Vector Embedding (RAG) - CAPA CAP backend-sample and fullstack-sample that parse an uploaded CSV of science data, embed each row through SAP AI Core and store the vectors in SAP HANA Cloud's vector store.sap official, cap
CAP samples embedding CSV rows into the SAP HANA Cloud vector store
Every CDS service in both samples is marked `@requires: 'any'` and left publicly accessible, which the README says simplifies the testing and deployment process, and it asks for proper authentication mechanisms to be applied in their place.
SAP BTP AI Best Practice Demo - Vector Query (RAG) - CAPSAP BTP AI Best Practice Demo - Vector Query (RAG) - CAPA CAP backend-sample and fullstack-sample that run a similarity query over the vectors in SAP HANA Cloud and pass the closest science-data rows to the model as context.sap official, cap
CAP samples running RAG similarity queries over SAP HANA Cloud vectors
Every CDS service in both samples is marked `@requires: 'any'` and left publicly accessible, which the README says simplifies the testing and deployment process, and it asks for proper authentication mechanisms to be applied in their place.
sap-btp-ai-best-practicessap-btp-ai-best-practicesSAP's collection of sample implementations for AI-powered applications on SAP BTP, one directory per topic under best-practices, in TypeScript, Python, Java and CAP.sap official, btp
Sample implementations for AI-powered applications on SAP BTP
SAP-samples/btp-cap-genai-ragSAP-samples/btp-cap-genai-ragA CAP reference application on SAP BTP for GenAI email support, built on SAP AI Core's Generative AI Hub and retrieval-augmented generation over SAP HANA Cloud's Vector Engine.sap official, cap
CAP reference app for GenAI email support with RAG on HANA Cloud
🟣 Bring AI into Your ABAP Applications with ISLM: Build, Integrate, and Consume AI Natively🟣 Bring AI into Your ABAP Applications with ISLM: Build, Integrate, and Consume AI NativelyDevtoberfest 2026 session AI-11, from the product manager and product owner of Intelligent Scenario Lifecycle Management: ISLM as the place an AI use case is registered, deployed and operated inside an SAP S/4HANA system, and what has been added to it for generative AI. The demo builds a custom scenario end to end — an intelligent scenario of type generative AI, a connection to SAP AI Core, a GPT-5 mini model, an orchestration flow template that configures data masking, input and output filtering and translation into the user's logged-in language, then prompts with a grounding query over five companies' return-policy documents — and consumes it from an ABAP report through the ABAP AI SDK, which calls the orchestration API and returns the summary as JSON. Two newer pieces follow: SAP RPT, the relational pre-trained transformer for tabular prediction, called with context rows and query rows and answering with a confidence value and no training step; and system-triggered agents, where an application job in the ABAP system starts a billing-creation agent that reaches S/4HANA OData APIs as tools through the agent gateway with principal propagation. The closing availability matrix states the generative AI capabilities for on-premise and private cloud from the 2021 release, with SAP notes to implement on lower ones.SAP Developers, published 2026-09-24
CAP in the age of AICAP in the age of AISeven live episodes of Hands-on SAP Dev in which DJ Adams works through CAP's AI features with the people who build them, each one a screen-share against a sample project. The first three are @cap-js/mcp: the @mcp annotation on a service, the describe, query and call tools, CQL as the query format, and what authorisation takes out of the tool list. From episode three the run moves to A2A and @cap-js/agents — human approval of an action, agents.md and skills, session persistence, telemetry and sub-agents — and the last two episodes are vector embeddings in CDS, local on SQLite and native on SAP HANA Cloud.published 2026-08-07
CAP in the age of AI #6 - vector embeddingsCAP in the age of AI #6 - vector embeddingsVector embeddings put to work in the incident management sample, answering whether this problem has been seen before. An embedding element of type Vector is calculated and stored from the vector_embedding function, so the database refreshes it whenever the text behind it changes, and cosine_similarity compares it with a freshly embedded string; because a title says little, a scheduled action first summarises each incident's conversation through a model on SAP AI Core and the embedding is taken from that summary. The similar-incidents list is a view joining incidents to themselves, with order and limit set in a before handler on the expanded query, and the vector element excluded from the service, since OData knows nothing about vectors — on SAP HANA Cloud the function is native, while SQLite needs packages added to the project.published 2026-10-09
CAP in the age of AI #7 - vector embeddings wrap-upCAP in the age of AI #7 - vector embeddings wrap-upThe same embeddings model taken from SQLite to SAP HANA Cloud, then handed to an agent. Locally it is @cap-js/ai that adds the vector functions SQLite lacks and downloads an all-MiniLM L6v2 ONNX model on first start; on HANA natural language processing has to be switched on, the model name is a mandatory argument, the vectors it returns are 768 dimensions, and dimensions must match on both sides of a comparison. HANA itself can be given SAP AI Core as a remote source for other embedding models, which a CDS.ql expression has to inject because the postfix notation only takes what the CDS model holds, and the hour ends with a search function over cosine similarity exposed over @cap-js/mcp, so an agent asked about low output from solar panels calls semantic search instead of writing the query itself.published 2026-10-16
arc-mcp-hubarc-mcp-hubA hub on SAP BTP that puts several arc-1 instances, one per SAP system, behind one login and a route each, exchanging the caller's token through a destination so each system still sees the real user.independent, btp
Hub on SAP BTP putting several arc-1 instances behind one login
Archived The declared repository is archived.
Functional boundary With one SAP system the README says to point the client at that arc-1 directly. Version 1 exchanges tokens within a single subaccount, so backends spread across subaccounts are out; it keeps its session map in memory, so more than one instance needs sticky sessions or a shared store. It routes deterministically and runs no model, which is a different thing from an assistant that reasons across systems. The `/all` endpoint drops the per-connection isolation, and the maintainer's answer to that is a PROD backend set read-only at the SAP end rather than a tool surface that hides the writes.
bw-modeling-mcpbw-modeling-mcpReads, creates and modifies BW/4HANA modelling objects through the same internal APIs Eclipse BWMT and the BW/4HANA Cockpit use — and can run centrally on BTP with each caller reaching BW as themselves.independent, btp
MCP server for BW/4HANA modelling objects
Its README says it plainly and so does this entry: this is not a simulation, every tool call reaches a live BW system, and write operations produce real changes. BW modelling objects are the definition of a warehouse — the blast radius of a wrong one is every report built on it.
cds-agentscds-agentsTurns a CAP service into a tool layer for agent frameworks — a CAPToolkit bound to a named service, with a tool strategy controlling how much of it becomes callable.independent, cap
Toolkit turning a CAP service into tools for agent frameworks
Maherd18/sap-mcp-serverMaherd18/sap-mcp-serverAn MCP server for ABAP development over the ADT resources: eight tools, each checked against a JSON policy file before the call reaches the system, with activation and transport release held for a person to approve.independent, abap
Policy-checked MCP server for ABAP development over ADT
Development and sandbox systems only: the README warns against pointing it at production. Transport release is never executed, since the tool exists to show the policy stopping it, and separation of duties is checked against the `identity.approvers` list in the policy file rather than against SAP authorisations, so it holds as far as that file does and no further. The write path has been tested with ABAP classes and other object types may need adjustments, there is deliberately no tool that creates or deletes an object, and one server instance serves one SAP system.
pyrfc-mcppyrfc-mcpAn MCP server for calling an SAP system's RFC-enabled function modules from an agent, built so that discovering a function does not authorise calling it: a read or a write is denied unless the function name matches an allowlist the server is configured with.independent, abap
MCP server calling RFC-enabled function modules behind an allowlist
Real SAP calls need the proprietary [SAP NetWeaver RFC SDK](external:sap-nw-rfc-sdk) and the `pyrfc==3.3.1` release the README calls archived and yanked, on Python 3.11, so without those only the demo backend answers. Not over HTTP as it stands: credentials are read from the environment for stdio, and the README says not to expose the server over HTTP without implementing the MCP authorization specification and deployment-level rate limiting. Multi-tenant auth and RBAC beyond the one server-wide policy, rate limiting, metrics and tracing, and deployment packaging are listed as not built. The result limits protect the client and not the system, since the README states they cannot reduce work already performed inside SAP, and the MCP tool annotations that describe read and write risk are advisory metadata rather than an authorisation boundary, so the service account's `S_RFC` and table authorisations are still what decides.
SuperClaude for SAPSuperClaude for SAPA Claude Code plugin that installs and wires its own ABAP MCP server, then puts a two-layer write guard between the agent and any system profile marked QA or production.independent, abap
Claude Code plugin with an ABAP MCP server and write guards
It is a plugin for one server — its own — so it is a stack rather than a component. And the guard is only as good as the profile labels: a production system registered as `DEV` is a production system with the safety off.
ABAP + MCP: Let AI Do Real Development in Your SAP SystemABAP + MCP: Let AI Do Real Development in Your SAP Systemabap-adt-mcp is an open-source MCP server that gives an AI agent full ABAP development capability over ADT: it searches, reads, edits, activates and unit-tests objects, and runs ATC checks, on a connected S/4HANA Cloud or on-premise system, acting as the signed-in developer. It runs on abap-adt-api, the same REST client Eclipse's own ADT tooling uses, and sits beside SAP's own ADT MCP Server, shipped with ADT for VS Code and Eclipse under a different server key, and marianfoo's abap-docs, a read-only server over the ABAP Keyword Documentation and Clean ABAP. Each destination carries a policy — read-only, allowed packages, denied tables and tools — the server checks before every call, and an optional audit file logs each tool call with its outcome and policy gate.Willian_Saez, published 2026-09-03
Agentic AI & AI AgentsAgentic AI & AI AgentsSAP's target picture for AI agents: Joule as the engagement layer — Joule Work, Joule Assistants, Joule Agents — with custom agents from customers, partners and SAP teams connecting through A2A for agent-to-agent collaboration and MCP for tool connectivity, on the SAP Business AI Platform. It says up front that parts are not generally available yet, in particular bidirectional communication with third-party agents through the Agent Gateway, so it reads as direction with the building blocks named rather than a design to copy today.published 2026-08-27
From APIs to Agent Tools: How MCP Gateway changes what Integration meansFrom APIs to Agent Tools: How MCP Gateway changes what Integration meansThe opening session of the Devtoberfest 2026 integration track, in which two product managers for SAP Integration Suite set out why the suite puts an MCP gateway between agents and backend systems, and then build one. The demo turns an HTTP endpoint and its OpenAPI specification into MCP tools on the integration cell runtime, does the same on top of an existing API-centric integration, and shows the policies tab where authentication through an external provider, quotas, spike arrest and IP filtering sit, plus the trace of the JSON-RPC calls in monitoring and the agent they are earmarked against in analytics. That is where the serve your application over MCP and tool governance edges hold; RFCs and BAPIs as sources are stated as supported, and MCP federation, custom domains and edge integration cell as roadmap.SAP Developers, published 2026-09-23
MCP Servers are the New Z-TablesMCP Servers are the New Z-TablesA single CAP service exposed as three overlapping MCP servers tools, queried by the same prompt nineteen times, returned the same tool every time — evidence, the author argues, that tool selection rests on the resemblance between a prompt and a tool's description rather than on any binding a where-used list could enumerate before retiring it. It contrasts that setup with @cap-js/mcp, which exposes a whole service as three generic tools instead of one per operation, and quotes the adapter's own beta documentation admitting no rate limiting, audit logging, input validation or approval workflow, and no use as a gateway to SAP Application APIs. From there it names four small Tool governance practices it has adopted: an owner recorded in the tool's own metadata, logging every invocation, deprecating inside the description, and exposing fewer tools.published 2026-08-21
Powering Enterprise-Ready Agentic AI with SAP Integration SuitePowering Enterprise-Ready Agentic AI with SAP Integration SuiteA Devtoberfest 2026 session (INT-31) from SAP Integration Suite product management on the MCP gateway. The demo creates an MCP server in a design package from an API already deployed on the tenant, exposes two of its 300-odd resources as tools, attaches authentication and quota policies on a policies tab, and publishes the deployed endpoint as an AI product in Developer Hub of API Management, where an agent subscribes and receives a token, URL, key and secret; the calls then show up in monitoring, in message processing logs and in the analytics dashboard, which is the MCP exposure and the tool governance the edges rest on. Three AI features inside Cloud Integration follow: a content migration agent that converts a PI/PO Java mapping into a Groovy script and generates a second integration flow for its lookup, memory profiling during a flow simulation, and root-cause analysis on a failed message processing log.SAP Developers, published 2026-10-07
Under no capability
Reports Assistant with SAP Document ManagementReports Assistant with SAP Document ManagementA CAP agent that turns a few notes into a status report and saves it in an SAP Document Management repository, where the instructions and the skills it follows are kept too.independent, cap
CAP agent writing reports into a shared SDM document workspace