Delivery Timeline

One milestone per consuming surface (the per-repo RFCs); Launchpad is the cross-repo anchor and the pilot. Effort is from rfcs/cross-repo.task-breakdown.md §1 Portfolio effort summary (all four repo totals reconcile exactly against their per-repo breakdowns). The BE here is each repo's FE-side current-company sync wiring (2 d × 4) — it excludes the upstream @mekari/sdk package/CDN + Session Manager + company-sync endpoint, which are owned by SSO/Account (BE), not Bifrost (shared blockers below).

Milestone	Source RFC	Effort (FE/BE/QA)	Confidence	Progress	Target Date	Status
Launchpad FE (pilot — Rollout step 4)	launchpad-fe	15.5 (FE 10 · BE 2 · QA 3.5)	low	0/15.5	TBD	Blocked (upstream)
CRM FE (step 5)	crm-fe	17 (FE 11.5 · BE 2 · QA 3.5)	low	0/17	TBD	Blocked (upstream)
Hub FE (step 5)	hub-fe	12.5 (FE 8 · BE 2 · QA 2.5)	medium	0/12.5	TBD	Blocked (upstream)
Hub Chat v2 FE (step 5)	hub-chat-v2-fe	13.5 (FE 7.5 · BE 2 · QA 4)	medium	0/13.5	TBD	Blocked (upstream)
Total	cross-repo	58.5 (FE 37 · BE 8 · QA 13.5)	low	0/58.5	TBD	—

Capacity & Allocation

Squad-wide view + the Load signal: bifrost/delivery/capacity.md.

Implementor	Discipline	Committed days (this initiative)	Window
Syafrizal M.	Frontend	37	TBD (gated on SDK availability)
Satya	Backend	8	TBD
Yoddi / Izzul	QA	13.5	TBD (end of build)

⚠️ Syafrizal M.'s 37 FE days on this initiative is the single largest line item in his quarter load, which now sits at exactly 1.00 (zero headroom) — see the roadmap Capacity summary / Program-level risks. Queue resequenced 2026-07-01 (EM decision): Qontak One Team FE (1.5 md) and Sales Invoice + Jurnal FE (2 md) now both run ahead of this initiative, pushing this FE's start back ~3.5 md. Moving this FE off Syafrizal M. (or adding a second FE) remains the open decision — see Dependencies & gates below for the resulting Q3-fit risk.

Dependencies & gates

Every repo is gated by upstream deliverables owned outside Bifrost — until they land, each repo can only build its mocked-SDK slice (cross-repo task-breakdown §2 shared blockers):

• 🔴 Account & Launchpad / SSO/Account (BE) — @mekari/sdk Session SDK package + CDN (SB-1, absent from all four repos' package.json); Session Manager service + Redis + Kong /sm/* routing (SB-2, no staging endpoint to integration-test against); current_company sync endpoint; canonical iframe path (/sm/current vs /sessionmanager/current); session.refresh() throttle interval (PRD "TBD").
• 🔴 infosec + Platform/SSO — CSP frame-ancestors whitelist mechanism; each product domain must be whitelisted SDK-side before its flag flips on at pilot.
• 🟡 SDK contract details — postMessage origin validation inside the SDK; SDK event-name contract (Hub OQ-4); hub-chat-v2 cross-domain _mekari_account readability (Q1, flagged structurally impossible if hub-chat must read the cookie).

Internal sequencing: Launchpad pilot (step 4) proves the SDK → CRM / Hub / Hub Chat v2 onboard (step 5). Within each repo: feature-flag chunk → SDK-loader chunk → plugin chunks; the SDK package (SB-1) + live Session Manager (SB-2) gate chunks 2–6 end-to-end in every repo.

• 🔴 QA E2E gate (Lane B) — finish-by date on the FE. This initiative is qa_lane: B: the 13.5-day E2E execution is a mandatory manual gate that runs after the integrated FE lands, serial on Yoddi (no automation substitute — Lane B). Design/automation can be authored ∥ the build against the RFC contracts, but the E2E window itself is full-duration and on the critical path. Derived constraint: for the 13.5-day E2E to finish inside Q3 (~W13), the integrated FE must complete by ~W10.3. The 2026-06-30/07-01 self-topup reconciliation freed 12.5 md off Syafrizal M.'s forward lane, moving this FE's projected finish to ~W9.7 — inside the ~W10.3 finish-by. Then, 2026-07-01 (later): the queue was resequenced (Qontak One Team FE 1.5 md + Sales Invoice + Jurnal FE 2 md now run ahead of this initiative, an EM decision to unblock both), pushing this FE's start back ~3.5 md and its projected finish to ~W10.6 — past the ~W10.3 threshold. E2E would then run ~W10.6 → ~W14.0, roughly a week into Q4. The FE-staffing decision (2nd FE / move SSO FE off Syafrizal M.) is therefore gated by a QA-driven finish-by date, not just FE hours, and is now the only lever that recovers this gate's Q3 fit. Surfaced as an at-risk filing in delivery/status/2026-06-30.md.

Notes

No hard external commitment date on this initiative (target_quarter: 2026-Q3, soft). The binding constraints are (1) SSO/Account delivering the SDK + Session Manager (no Bifrost task builds these; they gate every repo's pilot enable) and (2) the Syafrizal M. FE over-commit above. No RFC is execution-ready today — all four sit behind the four [critical] open contracts.