Feature request: first-class Product board (/product) in Grok Build host

[njivy]

Platform feedback: first-class product board for Grok Build

Audience: Grok Build / xAI harness team
From: User skill product-board (/product) — stopgap until host support
Date: 2026-07-27
Related host features: /goal, plan mode (/plan), /design, /implement, /execute-plan, /check-work, workflows, cross-session memory

Problem

Grok Build has strong execution and verification primitives:

Layer	Host / skill	Answers
Autonomous outcome	/goal + skeptic verification	Is this objective done with evidence?
Approach planning	Plan mode	How should we implement this?
Design → multi-PR	/design → /execute-plan	How do we ship a large change as a stack?
Impl ↔ review	/implement	Is this coding task clean?
Claim check	/check-work	Did this session actually deliver?

What is missing is a durable product portfolio layer:

• Ordered backlog across sessions (not todo_write, not chat memory alone)
• Milestone Definition of Done (multi-item ship bar)
• Capture of nits without forcing implementation
• Explicit dispatch from product intent → the correct execution tool
• Separation of “is the product ready?” from “did this goal run pass skeptics?”

Without that layer, users either (a) re-explain priorities every session, (b) overuse /goal for laundry lists, or (c) invent ad-hoc markdown that no agent consistently respects.

Proposal (host feature)

Add a first-class Product board (name flexible: Product, Backlog, Ship board) with:

1. Workspace-scoped board file (or structured store) with stable schema

   • Milestone + DoD checklist
   • Now / Next / Later / Icebox / Done
   • Items with goal-shaped fields: acceptance criteria, verification, non-goals, size (S/M/L routing only), dispatch hint

2. Slash surface e.g. /product status|capture|next|sprint|done|dispatch|init

   • Mirror the user skill contract so migration is trivial

3. Hard integration contracts (do not reimplement existing loops):

   • dispatch: goal → prefill /goal objective + AC/verification/non-goals into goal plan
   • dispatch: plan → enter plan mode with item as seed
   • dispatch: design / implement → hand off to those skills
   • Milestone completion ≠ goal completion; require DoD evidence aggregation

4. UI affordances (optional but high value):

   • Board panel or /product TUI view
   • Show active milestone DoD in session chrome
   • Link active /goal run to pb-NNN id

5. Non-goals for the platform feature

   • Story points, velocity, burndown, Jira sync (unless later demanded)
   • Replacing skeptic verification (stay on /goal)
   • Replacing plan mode

Why not “just a skill”?

Skills work (this repo proves the contract), but:

• Discovery and consistency across users is uneven
• No host-level link from board item → /goal plan artifact
• No first-class status when a goal is active vs board “Now”
• Platform incorporation enables shared schema, migration, and UI

The user skill should become a thin compatibility shim or be retired once the host feature ships.

Suggested acceptance criteria for v1 host feature

1. Board persists per workspace and reloads in a new session without chat context.
2. Item can be dispatched to /goal with AC/verification/non-goals intact in the goal plan.
3. /product done (or equivalent) refuses milestone completion when DoD checkboxes lack evidence.
4. Docs describe the layering: product board → goal | plan | design | implement.
5. Existing /goal / plan mode behavior unchanged when board unused.

Reference implementation

User skill path (local stopgap):

• ~/.grok/skills/product-board/SKILL.md
• ~/.grok/skills/product-board/references/board-template.md

Slash command today: /product (skill name: product).

Ask

Please consider incorporating a product portfolio / board primitive into Grok Build, wired to /goal and plan mode as above. Happy to iterate on schema or pilot with real projects (e.g. local-first app work using this skill).

---

Submission notes

• xai-org/grok-build has issues disabled, so this is filed on the official plugin marketplace tracker as the reachable public channel for xAI Grok Build-adjacent requests.
• A working user skill stopgap already exists locally (product-board, slash /product) and is available to package as an official marketplace skill if host support is deferred.
• Request priority: host integration with /goal + plan mode > packaging the skill alone.