The framework operates at the task level — how each work item is executed.
It plugs into Kanban, Scrum, SAFe, monorepos, and team Power Sessions without conflict.
Existing process organizes which work happens; the framework defines how it is executed
3-step lifecycle maps directly to Kanban columns: Context → Exploration → Production. Gates G1/G2/G3 become column transitions.
Flow management at the board level; task execution within each stage.
One Sprint Backlog Item = one lifecycle instance (CTX → EXP → PRD). G3 fulfils the Definition of Done.
Refinement produces better Context inputs — reducing CTX-Session duration.
Operates at the Team Level. Architectural Runway = ADRs from Context. Enabler stories stop after Exploration.
Does not replace ART coordination, portfolio management, or PI ceremonies.
The binding reference: load the framework via @url at the top of every project's CLAUDE.md.
Reads, greps and explanations are always free — the framework activates only when creating, editing, or deleting files.
Recommended directory layout for integrating into existing monorepos: context/, adr/, guidelines/, plans/, specs/.
Optional extension files for invariants, review criteria, testing strategy.
Synchronous 2-hour team sessions for cross-domain tasks. One session = one lifecycle step. Rule of thumb: needed when work requires knowledge from >1 person.
Default mode is 1 dev + AI, asynchronous — Power Sessions are the team-scale alternative.
Existing methodologies organize which work happens when. The framework defines how each task is executed. They operate at different levels.
You do not adopt the framework instead of Scrum or Kanban — you adopt it alongside them. Sprint planning, refinement, retrospectives, ART syncs, portfolio reviews all keep working as before. The framework only changes what happens between "task starts" and "task done".