Practitioner Playbooks
Ten procedures, each producing a specific artifact. Where a practice rests on measured evidence, the closing note says which; where it rests on reasoning because no published measurement exists, it says that too.
Ten procedures, each producing a specific artifact. Where a practice rests on measured evidence, the closing note of the chapter says which. Where it rests on reasoning because no published measurement exists, the chapter says that instead. Both kinds appear here, and the distinction is never blurred.
Designing an Oracle
Produces an oracle specification for one work class: what correct means, what decides it, and what the agent cannot see.
Loop Instrumentation
Produces an event schema, a small set of derived metrics, and thresholds calibrated from your own data.
Specifications Agents Can Build From
Produces a specification template and a worked conversion of one real requirement.
How to Stand Up a Work Class
Produces a registered work class with an owner, a demonstrated oracle, stop conditions, and a funded review budget.
The Context Substrate in Practice
Produces an entitlement-preserving retrieval layer, an indexed catalog, and an evaluated set of instruction files.
The Review Operating Model
Produces risk tiering, an auto-merge policy with a measured false-negative rate, and a seeded-defect probe.
Qualifying an Agent for Write Access
Produces a qualification record: what was tested, what it scored, who approved it, and when it re-qualifies.
Fleet Operations
Produces cost attribution, enforced quotas, tested kill switches, and telemetry whose analyzed fraction is known.
The First Ninety Days, Worked
Produces a sequenced ninety-day program with named artifacts rather than a list of intentions.
Beyond Ninety Days
Produces an operating plan for months four through eighteen, and the evidence Stage 3 cannot manufacture.