In-App Read-Only MCP Seam

Orchestrate in-process read-only MCP tool calls with a repo-owned registry.

Updated Mar 13, 2026
One-click install
npx skills add https://github.com/JDL440/nfl-eval --skill in-app-read-only-mcp-seam
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: In-App Read-Only MCP Seam
Source: https://github.com/JDL440/nfl-eval/tree/main/.squad/skills/in-app-readonly-mcp-seam
Command: npx skills add https://github.com/JDL440/nfl-eval --skill in-app-read-only-mcp-seam

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

In-app agents often need a safe, local surface to access tooling without leaking control to external providers. This Skill defines a repo-owned, read-only MCP seam that orchestrates tool calls inside the application runtime, using a single source of truth for the tool registry and strict guardrails.

Core Features & Use Cases

  • In-process orchestration: a runner coordinates proposals and tool executions against an allowlisted, repo-owned registry.
  • Read-only tool surface: restricts tools to non-mutating queries and data fetches, preventing publishing or side effects.
  • Reuse and guardrails: identical tool calls reuse cached results; violations are blocked; tests ensure registry alignment.

Quick Start

Set up the in-app MCP surface within your agent runtime to enable the read-only tool loop.

Frequently Asked Questions about In-App Read-Only MCP Seam

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

FAQPage Schema
How do I restrict in-app agents to read-only tool access?

To restrict in-app agents to read-only tool access, enforce a repo-owned tool registry that blocks mutating tools and limits the surface to safe queries and data fetches. This prevents publishing or side effects during runtime.

What is a read-only MCP seam for local tools?

A read-only MCP seam is a controlled interface that orchestrates local tool calls inside the application runtime. It uses a single source of truth for the tool registry to safely query data without leaking control to external providers.

How do I set up an in-process tool registry for agents?

To set up an in-process tool registry, define a repo-owned allowlist of non-mutating tools within your agent runtime. A local runner then coordinates proposals and executions against this registry to ensure safe orchestration.

Can I cache identical tool calls in an MCP tool surface?

Yes, you can cache identical tool calls in an MCP tool surface. The system reuses cached results for repeated queries, which optimizes local tool execution and ensures guardrails maintain registry alignment.

Does the read-only MCP seam block mutating tools?

Yes, the read-only MCP seam blocks mutating tools. It restricts the tool surface to non-mutating queries and data fetches, enforcing strict guardrails and tests to prevent side effects or unauthorized publishing.

When should I use a local MCP-style tool surface?

You should use a local MCP-style tool surface when in-app agents need safe, local access to tooling without leaking control to external providers. It is ideal for scenarios requiring validation and cached query reuse.