メニュー
Agent resolution table, working-scope declaration, and review mandates for the scaffold-dashboard orchestrator running inside a Docker container against the dashboard/ subtree of ue-claude-scaffold.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use for any container agent that posts progress to the coordination server's message board. Defines the curl format, channels, message types, payload conventions, posting responsibilities, and verbosity levels.
Use when a review agent must produce structured output. Defines the BLOCKING/NOTE/Summary/Verdict template, the two-tier confidence rubric, and the JSON shadow block consumed by POST /tasks/:id/reviews. Compose with review-process and a domain skill.
Use when an agent evaluates the output of other agents or decides whether work meets the bar. Enforces rigorous quality standards — reject shortcuts, demand evidence, no rubber-stamping.
Use when an agent is the FSM arbitrator on a durable task. Defines the two triggers, the three ruling types, the POST contract, the addendum-file rules for contradiction rulings, and the 409 handling. Project-agnostic.
Tells container agents where the Unreal Engine source tree is mounted inside the container, so they can grep and read engine headers to verify API signatures and class hierarchies.
Use when an agent must review code changes. Defines the universal review sequence — identify files, read context, check against domain criteria, score, filter. Compose with a domain skill and review-output-schema.
| name | scaffold-dashboard-system-wiring |
| description | Agent resolution table, working-scope declaration, and review mandates for the scaffold-dashboard orchestrator running inside a Docker container against the dashboard/ subtree of ue-claude-scaffold. |
| axis | environment |
Your working scope is dashboard/** — the dashboard subtree of ue-claude-scaffold. A task that requires changes under server/**, container/**, scripts/**, or anywhere outside the dashboard subtree is out-of-track; stop immediately and post phase_failed with scaffold-server-orchestrator named as the correct owner for server/** work.
Your sub-agents do NOT hardcode a scope — they rely on you to declare one in every delegation prompt. Every prompt you emit to the implementer, to any reviewer, or to the decomposition reviewer must include this line as its first directive:
Working scope:
dashboard/**. Refuse any task that requires touching files outside this scope, and flag any out-of-scope files in the changeset as BLOCKING.
Omitting the scope line is a protocol violation — the sub-agent will treat an unspecified scope as an error and refuse the work. This is how both the scaffold-dashboard track and any sibling dashboard orchestrator (e.g. content-catalogue-dashboard-orchestrator) safely share the same sub-agents without cross-contamination: each orchestrator declares its own scope, and the sub-agents trust that declaration.
| Role | Agent | Purpose |
|---|---|---|
implementer | scaffold-dashboard-implementer | React, Mantine, TanStack Router/Query; TDD via Vitest |
style-reviewer | scaffold-dashboard-react-quality-reviewer | Component discipline and TypeScript style opinions fused |
safety-reviewer | scaffold-dashboard-browser-safety-reviewer | React-agnostic web hygiene (XSS, URL, CSRF, storage) |
reviewer | scaffold-dashboard-correctness-reviewer | Spec compliance, cache invalidation, project scoping, test coverage gate |
decomp-reviewer | scaffold-dashboard-decomposition-reviewer | File bloat, hook decomposition, route layering |
Each reviewer assesses only its own dimension. The style-reviewer slot is filled by the react-quality reviewer — there is no separate style reviewer and no separate type reviewer for the dashboard track, because React ESLint already covers the mechanical style surface and TypeScript type presentation is part of component discipline. The safety-reviewer slot is filled by a browser-safety reviewer that is deliberately React-agnostic — it reviews what would render in any browser runtime, not framework-specific render-stability bugs.
These agents have dashboard conventions and enforcement baked into their definitions. Your delegation prompts should focus on what to do (the phase requirements, file lists, specification), not how to work (build commands, style rules, environment details).
The phase protocol skill hardcodes [STYLE REVIEW] and [SAFETY REVIEW] tag strings for Step 2c message posts. For the dashboard track you must override these tags when posting reviewer output to the message board:
scaffold-dashboard-react-quality-reviewer → tagged [REACT QUALITY REVIEW] (not [STYLE REVIEW])scaffold-dashboard-browser-safety-reviewer → tagged [BROWSER SAFETY REVIEW] (not [SAFETY REVIEW])scaffold-dashboard-correctness-reviewer → tagged [CORRECTNESS REVIEW] (unchanged)The reviewers themselves still emit output in the standard review-output-schema format; only the tag on the message-board post changes, so the operator reads an accurate label.
Step 2 of the phase protocol runs three parallel Agent calls — the three roles above — all with review-output-schema verdicts. All three must APPROVE for the phase to pass. There is no fourth reviewer.
The dashboard track has no dedicated tester agent. The implementer writes Vitest tests as part of its TDD loop. The correctness reviewer enforces a test coverage gate — any new hook, util, or component-extracted logic path without a corresponding Vitest test in the same commit is BLOCKING. Do not attempt to delegate to a tester; there is no such role in this wiring table.