메뉴
Scaffold and govern an MCP server as a bounded tool seam. Defines which tools the agent may call, enforces the principle that agents call tools and tools call infrastructure. Trigger when adding or reviewing an MCP server in an agent system.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Bootstrap or harden a project's operating model for mixed human and AI delivery teams. Use when starting a repository, adopting an agent harness, defining authority and risk tiers, coordinating parallel agents safely, binding evidence and review to an exact candidate, or correcting fragmented rules that prevent coherent, consistent, complete delivery.
Architecture decisions, trade-off analysis, and ADR authoring. Trigger when a significant technical decision needs to be made, documented, and owned — not when an implementation detail needs a choice.
Decompose an epic into atomic parallelizable tasks and route each task to the right skill. Use this as a meta-router when you have more than one skill available and need to decide which applies. Trigger at the start of any multi-track epic or when the skill count in your harness exceeds ~12.
Check proposed stack, data flows, cloud choices, and delivery controls against declared enterprise policies, compliance frameworks, and security controls. Trigger at R2/R3 risk classification, before locking architecture decisions, or when the operating profile names a governance pointer.
Govern the SemVer release process — versioning discipline, tag-based GitHub releases, changelog hygiene, and the ADR that confirms the release strategy is agreed. Trigger when setting up a release process, before a first public release, or when release practice has become inconsistent. Owned by delivery-orchestrator.
Session-start orientation for this team harness. Load first, before any work. Explains the operating contract, when each skill triggers, and how installed companion plugins supply agent-craft capabilities while this harness supplies the team contract. Trigger at the start of every session and whenever unsure which skill applies.
| name | mcp-server-scaffold |
| description | Scaffold and govern an MCP server as a bounded tool seam. Defines which tools the agent may call, enforces the principle that agents call tools and tools call infrastructure. Trigger when adding or reviewing an MCP server in an agent system. |
Agents call tools. Tools call infrastructure. This boundary is the governance seam. Do not collapse it.
An MCP server is a contract between an agent and the capabilities it is allowed to use. Its value is not in what it exposes — it is in what it deliberately withholds.
Define the tool seam — before writing any server code, list:
Author the tool manifest — for each tool:
Tool: <name>
Description: <one sentence — what it does and when to call it>
Input schema: <typed fields with constraints>
Output schema: <typed fields>
Side effects: <what it changes in the world — none / read-only / write / external-call>
Auth required: <yes/no and mechanism>
Rate limit: <calls/minute or none>
Enforce least-privilege at the server — the MCP server holds the credentials and enforces the scope. The agent receives only the tool interface. Key rules:
Add input validation at the boundary — every tool validates its inputs before executing. Reject malformed inputs with a structured error, not an exception trace.
Log every tool invocation — at minimum: tool name, caller identity, input summary (no secrets), output summary, timestamp, and latency. These logs are your audit trail.
Test the refusal surface — confirm that:
Document the scope boundary — in the MCP server's README or SKILL.md, explicitly state what the server does NOT expose and why. Scope boundaries that are not documented drift.
Run the Adversarial Gate — common MCP failure modes: overly broad tool descriptions that invite misuse, missing input validation that allows prompt injection via tool arguments, write tools with no confirmation step, no rate limiting on expensive operations.