CAP
SAP's cloud application programming model for building enterprise extension applications, supporting development in multiple languages and intended to achieve clean core extensibility.
Timeline (last 90 days)
In development mode, CAP's n8n plugin replaces workflow triggers with log output, allowing local development without a real n8n instance; asynchronous triggering ensures consistency through the transactional Outbox pattern.
Developers can place an agent.md file alongside a CAP service to customize agent behavior, such as setting the tone or adding skill descriptions.
CAP's early MCP adapter used CQN (a query representation in the form of JSON objects) and later switched to a SQL-like textual query language called CQL, and the option to switch back to CQN has been removed.
CQL syntax is highly similar to SQL SELECT statements, supporting where, order by, join, aggregation and more, while retaining CAP's shorthand support for association paths; its SQL-like form lowers the barrier for large models to generate it and improves call success rates.
CAP's n8n integration plugin can, via annotations or custom logic, trigger n8n Webhooks when events such as entity creation occur, invoking integration capabilities in the n8n ecosystem.
CAP's AI-related plugins exist as standalone npm packages under the CAPJS namespace, and developers can install them as needed.
Adding the @mcp annotation to a CDS service exposes the existing service to external agents via the MCP protocol.
CAP's MCP adapter exposes only three generic tools per service: describe, query and call, to avoid blowing up the large model's context window with a large number of tool definitions.
The MCP adapter's tool descriptions come directly from doc comments in the CDS model; ordinary comments are not injected.
Adding the @agent annotation to a CAP application gives the application an agent interface that can be conversed with in natural language.
4 more timeline events
All Facts (15)
CAP是由国内开发者杨晓东(Savorboard)创建的基于.NET的分布式事务解决方案库
80%UnverifiedIn development mode, CAP's n8n plugin replaces workflow triggers with log output, allowing local development without a real n8n instance; asynchronous triggering ensures consistency through the transactional Outbox pattern.
50%UnverifiedDevelopers can place an agent.md file alongside a CAP service to customize agent behavior, such as setting the tone or adding skill descriptions.
50%UnverifiedCAP's early MCP adapter used CQN (a query representation in the form of JSON objects) and later switched to a SQL-like textual query language called CQL, and the option to switch back to CQN has been removed.
50%UnverifiedCQL syntax is highly similar to SQL SELECT statements, supporting where, order by, join, aggregation and more, while retaining CAP's shorthand support for association paths; its SQL-like form lowers the barrier for large models to generate it and improves call success rates.
50%UnverifiedCAP's n8n integration plugin can, via annotations or custom logic, trigger n8n Webhooks when events such as entity creation occur, invoking integration capabilities in the n8n ecosystem.
50%UnverifiedCAP's AI-related plugins exist as standalone npm packages under the CAPJS namespace, and developers can install them as needed.
50%UnverifiedAdding the @mcp annotation to a CDS service exposes the existing service to external agents via the MCP protocol.
50%UnverifiedCAP's MCP adapter exposes only three generic tools per service: describe, query and call, to avoid blowing up the large model's context window with a large number of tool definitions.
50%UnverifiedThe MCP adapter's tool descriptions come directly from doc comments in the CDS model; ordinary comments are not injected.
50%UnverifiedAdding the @agent annotation to a CAP application gives the application an agent interface that can be conversed with in natural language.
50%UnverifiedUpgrading from classic ABAP skills to skills such as BAS, CAP, Cloud Foundry, and SAPUI5 requires 6 to 9 months of sustained investment, not the 8 weeks claimed by the vendor.
50%UnverifiedIRCv3's capability negotiation is implemented through the CAP subsystem: the client sends CAP LS to obtain a list of features, uses CAP REQ to request enabling capabilities, and the server responds with CAP ACK or CAP NAK.
50%UnverifiedThe CAP capability negotiation process occurs before the formal login; legacy clients that do not support the CAP command will ignore these interactions, and the server falls back to compatibility mode, allowing new and old clients to coexist in the same channel.
50%UnverifiedCAP trade-off refers to making choices among consistency, availability, and partition tolerance.
50%