API documentation

Architecture pattern. K Pro exposes two programmable surfaces: a REST API for synchronous job submission, status polling, and result retrieval (structured JSON with provenance), and an MCP server endpoint that lets external agents discover and invoke K Pro platform directly.

Long-running work (e.g. literature mining, multi-omic queries, federated analyses) runs asynchronously through the orchestrator. Submissions return a job ID; completion may be delivered via webhook callback or polled status.

Representative use cases.

• Batch target scoring. A differential-expression pipeline submits a gene list; the orchestrator runs the target-evaluation agent over each entry and returns a ranked JSON payload with confidence and evidence links.

• Pipeline step. Nextflow or Snakemake stages call K Pro as a typed REST node, blocking on a webhook callback before downstream steps proceed.

• Embedded reasoning. A customer's own agent calls K Pro tools over MCP to compose target-evaluation reasoning into its own workflow without round-tripping data through a UI.