Organizations are increasingly encoding operational procedures as multi‑step agent skills, called Agent Operating Procedures (AOPs). These procedures must be traceable, auditable, reproducible, access‑controlled and bounded in time and cost, especially in regulated sectors. Traditional AOPs are kept as coarse files where steps bundle actions, copied procedures diverge, any step may call any tool, and refactoring is unchecked.\ \ Functional Agent Operating Procedures (FAOP) bring database‑grade discipline to AI agent workflows. FAOP treats each workflow as a typed function over scoped state and subjects an entire library to four progressive normal forms:\ \
- Decomposed (FNF‑1) splits a workflow into atomic steps;\
- Typed (FNF‑2) makes data flow explicit and type‑checked;\
- Factored (FNF‑3) gives each shared procedure a single canonical definition so changes propagate instead of drifting;\
- Scoped (FNF‑4) confines every step to least‑privilege tool access.\ \ After normalization, every atomic step is a separate entity with a declared contract, allowing independent type checking, scope confinement, bounding, testing, tracing and auditing. A FAOP algebra of provably sound equalities turns refactoring and parallelization into compiler rewrites. Orthogonal extensions add verified termination (+V), optimization (+O) and time‑and‑cost bounds (+B). An architectural classifier (0‑3) decides whether a workflow truly needs AI. Together these make integrity, isolation, typing, termination, cost and optimization compile‑time concerns rather than runtime hopes.\ \ We implement the theory in SkillBase, an AOP management system analogous to PostgreSQL for data, together with a separate executor that enforces the guarantees at runtime. A worked financial workflow illustrates the end‑to‑end approach.\ \ Review