Updates

QA and regression complete on the §4.5 legacy platform changes; no open blocking defects.

Unit, integration, E2E, accessibility, performance, and regression coverage per §6 complete across rad-helix and rad-platform-layer.

The registration dispatcher is the highest-risk code in the app and has no test. Cover the allowlist (unlisted path, wrong method, undeclared query param all rejected before upstream), the CSRF pair (minted per mutation, Set-Cookie relayed to the browser), the provisioning timeout, and pass-through fidelity — a TrialStepResponse must reach the client unmodified, including the statuses that represent refusals.

Exit criteria: every ENDPOINTS row exercised against a mocked trial_manager; a 200-with- refusal-status never surfaces as an error, and a transport fault never surfaces as a step.

Playwright walk of the funnel in each state the step machine can put a visitor in: new email, existing account, SSO-required, student-blocked, expired campaign, out-of-region, plus trial-one and trial-everything. Demo mode drives the signed-in/signed-out split.

Exit criteria: each branch reaches a correct terminal screen; no branch dead-ends or shows a raw error.

Opt-in banner, user setting, and nav changes shipped to production ahead of go-live.

(RC-1110, 5 pts, To Do)

There is no test harness in the repo yet. Scaffold Vitest + @nuxt/test-utils and cover the campaign and catalog routes against mocked upstreams: session forwarding per environment, write-flag gating (405), no-session behavior (401), and upstream-error mapping. These gates are the newest logic in the app and the easiest to regress silently.

Exit criteria: suite runs in CI with no live upstream; every write route asserts its 405 and 401 paths.

(part of RC-1125)

Playwright walk-through of the sidebar shell, campaign create/edit/archive/publish, catalog create and edit across all three record kinds, contact creation, and Trends/Outreach against a mocked analytics payload — including the write-disabled variants of each screen.

Exit criteria: every primary flow green in both write-enabled and write-disabled configurations.

(part of RC-1125)

The routing exists but has not carried a real deployment. Merge to develop, confirm the dev stack builds and deploys, and confirm the app authenticates against the dev Trial Manager and the content.dev session cookie — the one path where mixing DE_ENVIRONMENT and trialManagerEnv would drop the session.

Exit criteria: dev URL loads authenticated, campaigns and catalog read live dev data.

All P1 must-haves at Built; go/no-go checkpoint for Aug 30 launch.

LIMITED - teacher opt-in testing available; capability to activate for a small subset of users the same day.

The funnel now depends on trial_manager for every branch, so the deploy matters more than usual: verify against the dev stack that the anonymous CSRF pair survives the real cookie domain, that provisioning completes inside its timeout, and that the fallback paths don't mask a healthy backend.

Exit criteria: a real trial provisioned end to end in dev with no local heuristics involved.

(part of RC-1125)

Document the environment flags (TRIAL_MANAGER_ALLOW_WRITES, CONTENT_API_ALLOW_WRITES, DE_ENVIRONMENT, trialManagerEnv), the per-environment session-cookie model, and the data-gap workflow, then decide the write-flag posture per environment before prod.

Exit criteria: a developer new to the repo can run it locally, explain the flags, and add a data gap unaided.

Close the five open district confirmations and put the Grade 4 Q1 guide in front of the teachers who would run it.

• Confirm CKLA unit order by quarter, Ready Math pacing, and whether Mystery Science unit order is actually flexible.

• Confirm arts specialist involvement and whether collaborative planning time exists.

• Name the pilot grade and teacher team for the coming year.

• Write the definition of done for an arc — the rubric that later becomes the automated validator set.

• Exit criteria
  A teacher team walks the Q1 arc unaided and says they could run it, plus a written arc rubric.

Refinements and fixes shipped based on telemetry, Pulse feedback, and further QA from the initial opt-in cohort.

Opt-in expanded to the Phase 2 cap

Refinements and fixes shipped based on Phase 2 cohort feedback and QA.

Opt-in expanded to the Phase 3 cap; scale/performance validation per §6 confirmed ahead of the 20x jump.

Refinements and fixes shipped based on Phase 3 cohort feedback and QA.

Opt-in expanded to the Phase 4 cap — final expansion within this plan's window.

Tentatively use in Q4 to complement PI Planning (led by @Nikkita Willis). Dates TBD.

The single highest-leverage step. Convert the prose design principles into a machine-readable model so that generating an arc becomes filling a structure, not writing a document.

• Define the arc data model — district, program, pacing week, standard, concept, subject thread, touchpoint, alignment record, culminating product.

• Build the two output templates as renders of that model: leadership overview and teacher implementation guide.

• Encode the rubric from Phase 0 as executable validators.

Exit criteria The Q2 prototype is re-expressed as structured data and re-rendered from it with no loss of quality.

Build all of K–5 for Jefferson County — roughly 24 arcs — with human experts on the schema and light tooling. Instrument the work from arc #1 so the automation targets are measured, not guessed.

• Track hours per arc, edits per arc, and which steps consume the time (standards reading, overlap hunting, asset search, or writing).

• Harvest reusable pieces: standards overlap maps, per-program pacing normalizations, the culminating product menu.

• Ship arcs continuously to the pilot so instructional feedback arrives during the build, not after.

Exit criteria
Median build time per arc down 60%+ from arc #1 to arc #24, with a time breakdown that names the bottleneck.

Build the engine that does the mechanical work: reading pacing guides and standards, detecting genuine overlap, and attaching the right DE assets to the right week. Detailed in the next section.

Exit criteria
≥85% of proposed alignments accepted with no edit; expert review under two hours per arc.

Everything district-specific becomes configuration data. Nothing about Alabama or Mystery Science is hard-coded anywhere.

• Curriculum program library: each publisher’s scope and sequence normalized once, versioned, and reused by every district that licenses it. Onboarding district #2 with the same programs costs nothing.

• Standards crosswalk: every state’s standard set resolved to the same canonical concept layer, so overlap detection works in any state without pairwise mapping.

• Proof: a second state with a different publisher stack, built with no code changes.

Exit criteriaA new district is live in ≤2 weeks with zero engineering involvement.

Turn the internal pipeline into a surface other people can drive.

• Authoring workbench opened to district curriculum staff, not just DE experts.

• Teacher delivery experience inside DE — the arc as a live, dated plan rather than a PDF.

• A library of published arcs, filterable by state, publisher stack, and grade.

• Usage telemetry and teacher edits feeding back into alignment scoring.

Exit criteria
A district authors a usable arc without DE in the room.

Concept prototype, tested with a few trusted partners. Confirm problem, solution, and willingness to pay.

Working MVP with 1–2 district-approved AI platforms. All DE content indexed and searchable via MCP (DEX, Mystery Science, Techbook, CEP).

Test the trust value prop with a small set of pilot districts already on DE.

Size investment for broader Connected Ecosystem rollout.