board

Generates interactive HTML review panels for requirements and design documents with feedback write-back.

Updated Jan 12, 2026
One-click install
npx skills add https://github.com/ab300819/keel-workflow --skill board-ab300819
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: board
Source: https://github.com/ab300819/keel-workflow/tree/main/skills/board
Command: npx skills add https://github.com/ab300819/keel-workflow --skill board-ab300819

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Reviewing dense requirements (01-requirements.md) and system design (02-system-design.md) documents in raw Markdown is hard for humans, and collecting structured feedback that flows back into those documents without breaking numbering or traceability is error-prone. This Skill renders the documents as a self-contained HTML review board where users mark items pass/reject/hold with comments, then converts that feedback into controlled document revisions. ## Core Features & Use Cases - Interactive Review Board: Parses 01/02 documents into per-item review cards (F/US/AC/ADR anchors, interfaces, modules, sections) with a traceability tree and coverage map, opened in the user's real browser via chrome-devtools MCP. - Closed-Loop Feedback Write-Back: Reads back the review state, converts each reject/hold comment into concrete document edits, confirms them with the user, and writes revisions while preserving IDs, change history, and the F→US→AC trace chain. - Safety Discipline: Fail-closed board_id validation on every script, read-only access to window.__review, HTML-safe JSON injection, network-isolated mermaid rendering, and a T2 export/paste fallback when MCP is unavailable. - Use Case: After running /requirements and /system-design, run /board to let stakeholders review acceptance criteria visually, then have the agent apply their decisions back into the documents and register conclusions in the design review table. ## Quick Start Run /board after generating the 01 requirements and 02 design documents to open an interactive review page and feed the collected verdicts back into the documents.

Frequently Asked Questions about board

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review requirements and design documents interactively?

Run /board after generating 01-requirements.md and 02-system-design.md. It parses the documents into per-item review cards in a self-contained HTML page where you mark each item pass, reject, or hold with comments, then the agent converts your feedback into document revisions.

How does review feedback get written back into documents?

Each reject or hold comment is converted into a concrete edit proposal and confirmed with you before writing. Existing item numbers stay unchanged, revisions are logged in the change history column, deletions are marked deprecated rather than removed, and design changes go through incremental design with ADRs.

Does the board skill work without chrome-devtools MCP?

Yes, it falls back to a T2 transfer mode. The page opens with the system browser and an export button serializes the review state as JSON, which you paste back into the conversation for strict schema and hash validation before import.

Can I use the board skill to review code or PRDs?

No, it is scoped to keel 01 requirements and 02 system design documents only. Code review and PRD review are explicitly out of scope; for external LLM-based design critique, the adversarial-review skill is the complementary option.

Why does mermaid diagram rendering fall back to source code blocks?

Mermaid source is treated as untrusted input, so rendering requires network deny-all isolation such as macOS sandbox-exec. When isolation is unavailable, the board renders the diagram as a source code block instead of executing it.