← Back to the BioCreative OS BioCreative

Methodology Map

How we decide what to build and what to run it on: the eight-step intake, the loop/swarm/harness vocabulary, and the full harness menu with production status and lock-in cost.

The build shape, before the framework

Every capability gets these eight decisions, in this order. Skipping to "which framework" is how you end up with a LangGraph deployment for a task that was a single structured completion.

#DecisionDefault
1Processor — one-shot, loop, or agent?one-shot
2Harness — what environment runs it?Pydantic AI / brain-api pattern
3Shape — loop-shape × swarm-shapeone-shot × single
4Skills — which of the 219 does it load?the fewest that close the task
5Context — what must it know, per tenant?MCS spine, projected
6Memory — what may it write back, and how?propose-then-confirm
7Eval — how do we know it worked?a stated floor, never "looks right"
8Trigger — what starts it?manual until it earns a schedule
Loop is depth. Swarm is width. Harness is the box. They are three independent axes and conflating them is the most common planning error. Google's 2026 data: swarms help genuinely parallel work ~80% and degrade sequential reasoning 39–70%. A single agent tops out around 10–15 tools.
Any shape other than one-shot needs codified breakers before it runs: max_turns, a per-run dollar cap, duplicate-call detection, a wall-clock timeout, and an explicit "I'm stuck" exit. Measured cost of skipping them: one self-prompting run exceeded 1M tokens; one managed-agent run burned ~$12 in 28 minutes and still failed its rubric.

The harness menu

Patterns you can build on — deliberately not BC's own products. What BioCreative actually shipped and where it runs lives in the Agent Fleet Registry. Lock-in is scored 1–5: how hard it would be to leave.

25
harnesses in the menu
15
in production
27
deep docs
80
agents in the fleet