Build the L2 "adoption readout" generator skill (sibling of generate-landing — MCP-retrieval harness + KNOWLEDGE-BRIEF; a 4-axis scorer over process/workflow atoms → sequenced roadmap → OKF artifact in the client repo)

task-l2-adoption-readout-generator

task confidence verified status done 2026-06-18 owner knowledge-architect

source log-auditor — surfaced recording 0052-agentic-agency-in-a-box-and-adoption-ladder. DONE 2026-06-20 (FDE, per the product-owner spec DEC-0067): the deterministic 4-axis scorer ships in packages/okf-view/src/adoption-readout.ts (9/9 tests), the harness in scripts/adoption-readout.mjs, and the .claude/skills/adoption-readout skill narrates + emits the OKF artifact. Verified end-to-end on the real RBA tenant — 11 initiatives scored (all inferred, honest Now/Later split), and a real readout artifact emitted to clients/rba/readouts/ validates via @dossier/okf parse() (type=artifact, confidence=inferred=floor), with all 11 relates_to/[[links]] resolving and zero fabricated $/%; Dossier KB untouched (isolation held).

Build the L2 "adoption readout" generator skill

One loop, three faces — ship an agentic Digital Experience Agency "in a box" (the Agency) plus an adoption-ladder readout product (the Ladder) over the OKF Company Brain, dogfood-first's third face — the Ladder — is the L1→L14 adoption onramp / diagnostic readout computed from the org's OKF Company Brain. It is the wedge, and a seed already exists as the per-client KNOWLEDGE-BRIEF.md. This task builds the shippable L2 readout generator.

The work

A skill — the sibling of generate-landing — that:

Reads the client's own OKF via the same MCP-retrieval harness (@dossier/mcp / readKbAtoms) + the KNOWLEDGE-BRIEF.md seed.
Runs a 4-axis scorer over the org's process / workflow atoms — the four axes operationalize Alex Lieberman's L2 prioritization (e.g. ROI / risk / cultural support / readiness; pin the exact four in the skill spec).
Produces a sequenced roadmap (the L2 post-diagnostic readout: a transformation timeline by initiative).
Emits it as an OKF artifact in the client repo — sovereign, owned, re-derivable (Adopt OKF as Dossier's canonical knowledge format: the readout lives in the client's own git, not a Dossier-held report), linking (never copying) the atoms it scored.

Like generate-landing, it reads the client's own OKF and writes a tailored artifact into the client's repo, never perturbing Dossier's own KB.

Why a task, not a fix-in-place

A net-new skill with a scorer, a roadmap shape, and an OKF-artifact emitter — real Principal Knowledge-Format Architect work (the readout is an OKF concept the model owns), not a hygiene fix. confidence: inferred (agent-filed from DEC-0052's named follow-up).

Coordinate, don't duplicate

This is a new skill, the sibling of generate-landing (whose own stale-path follow-up Fix the stale round-trip-test path in the generate-landing SKILL.md (points at a file that no longer exists) is already done). Board globbed before filing: no open task covered an adoption readout / L2 diagnostic / ladder scorer.