To make the integration of the sproof Sign API as efficient as possible, you can configure a specialized AI agent. This acts as a virtual Senior Solutions Architect and assists you in choosing the right integration strategy as well as the correct use of the endpoints.
Configuration of your AI agent
To create a reliable assistant that gives precise architectural recommendations without hallucinating code, you must equip it with specific knowledge and clear behavioral rules. The focus here is on the four basic integration approaches of sproof Sign.
Proceed as follows to set it up:
-
System prompt: Copy the contents of the file sproof-system-prompt-V4.md into your agent's system instructions. This establishes that the agent responds exclusively based on the V4 sources and never generates code examples.
-
Upload knowledge resources: Upload the following documents as a static knowledge base (resources) into your agent so that it accesses only verified information:
-
Document 1: sproof-canonical-request-structures-V4.md
-
Document 2: sproof-customer-qa-V4.md
-
-
Validation: Ensure that the agent follows the strict grounding rules. It must not invent endpoints or parameters that are not contained in the uploaded documents.
Important behavioral rules for the agent
The agent has been instructed that in cases of uncertainty or missing information in the source documents it should always indicate that there is not enough information available, rather than guessing an answer. Its primary purpose is to identify the appropriate one of the 4 integration approaches for your scenario.