Registering an agent

agent registration captures the agent’s metadata and intent this step does not include credentials or runtime configuration registration captures metadata only runtime configuration (llm, connections, and credentials) is defined later during deployment interaction modes you can choose to integrate agents that are native to the platform or use external ones it also helps you manage, govern, and clearly find and execute your agents define the interaction model of the agent conversational agents designed for user interaction via prompts/questions non‑conversational agents created for background workflows, automation, or api triggered execution; this is not directly accessible to end users in conversational interfaces define where the agent executes bridge hosted agents external hosted agents (executed via external apis like watsonx or others) registering and agent from an existing repository upload a file for bulk agent registration use file upload for registering multiple agents at once locate the manifest json file within your repository this file typically contains essential metadata and configuration details for the agent you wish to import source github kyndryl agentic ai org → bdg sw agents base repo → config/ directory review the file and update account specific fields tenant id — the account's bridge tenant id connection ids — map to the connections created access group id — map to the access group service id — map to the service id knowledge base id — map to the knowledge base (if applicable) llm api key id — map to the api key once the information is correct, go to the agent library page, and select register select the file upload option select the manifest json file from your local system click upload after uploading, the system will process the file, extracting the necessary information to integrate the agent into the library seamlessly manual registration single registration is the manual process to register an agent from the agent library page, select register select single registration basic information type select agent or agentic solution name enter a unique name for the agent version enter the version number description add a short description owner enter the agent owner maintaining team enter the team responsible for updates tags add tags separated by a comma technology framework select the framework used summary enter a summary of the agent’s purpose select next to continue skills skill name enter a unique skill name skill description add a short description select add skill to add more or next to continue configurations llm configuration select llm configuration llm model select model knowledge base configuration select global or private knowledge base select a knowledge base mcp mcp server select mcp server mcp tool select mcp tool connector type select connector type select next to continue schema enter input and output schema or select default select next to continue source of agent github repository url enter github repository url branch enter repository branch manifest path enter manifest path select next to continue flow configuration agentic solution only enter json or yaml flow select next to continue review and submit review all information and select submit