Series
Nine patterns for building your second agentic product without forking your first — extracted from building HUGO, an internal multi-product agent platform built on LangGraph primitives.
Start here
Extracting a traditional library is mostly mechanical — you can read the imports. Agentic systems hide three more coupling vectors where your toolkit can't see them: in prompts, in state field names, in the rules the planner enforces. The second product doesn't take a fraction of the runway because of effort. It takes it because those abstractions don't exist yet.
· 18 min read
The function that materializes your graph can have zero application imports. Topology becomes a declarative record the application supplies and the platform composes unchanged.
· 22 min read
A tool's entire interface with the orchestrator — schema, config, invariants, state reads and effects — collapses onto one declarative record, and the registry becomes the dispatcher.
· 16 min read
The rules that govern what counts as a valid plan belong on an injected policy the application supplies, not in the planner's prompt or your core — so the answer can differ per product without forking the planner.
· 15 min read
Name your model-and-params combinations as profile records the application registers at bootstrap, and let node code ask for a model by role so configuration scales with the graph instead of against it.
· 16 min read
Load skills from a directory with role-aware conflict resolution, and assemble their outputs into the response through a registry the core reads — so adding a capability means adding a directory, not teaching a core node another skill's fields.
· 28 min read
The two smallest, most-reused mechanisms in the catalog — a tiered config resolution order with normalizers, and a prompt-slot injection seam that routes one call's output into the next call's prompt without anyone wiring it up.
· 21 min read
The same registry that decides what a turn writes into its response can drive every read path too, so the write model and its read models can't drift apart.
· 12 min read
Who owns the field names the LLM produces — and why getting them out of your framework takes two patterns, not one.
· 13 min read