Agent-Skills.md

Agent Skills: Skill: mcp-toolkit

Unified toolkit for MCP (Model Context Protocol) operations including server building, access validation, extension verification, message board management, pre-commit checks, and smoke testing.

UncategorizedID: Reneromero08/agent-governance-system/mcp-toolkit

Repository

Reneromero08/agent-governance-system
Reneromero08License: NOASSERTION

Install this agent skill to your local

pnpm dlx add-skill https://github.com/Reneromero08/agent-governance-system/tree/HEAD/CAPABILITY/SKILLS/mcp-toolkit

Skill Files

Browse the full folder contents for mcp-toolkit.

Download Skill

Loading file tree…

CAPABILITY/SKILLS/mcp-toolkit/SKILL.md

Skill Metadata

Name
mcp-toolkit
Description
"Unified toolkit for MCP (Model Context Protocol) operations including server building, access validation, extension verification, message board management, pre-commit checks, and smoke testing."

required_canon_version: >=3.0.0

Skill: mcp-toolkit

Version: 1.0.0

Status: Active

Purpose

Unified toolkit for MCP (Model Context Protocol) operations including server building, access validation, extension verification, message board management, pre-commit checks, and smoke testing.

Trigger

Use when performing any MCP-related operation:

  • Building MCP servers
  • Validating agent access patterns
  • Verifying MCP extensions in IDEs
  • Managing message board operations
  • Running pre-commit MCP checks
  • Performing MCP smoke tests
  • Wrapping MCP server execution

Operations

| Operation | Description | |-----------|-------------| | build | Build MCP servers (placeholder) | | validate_access | Validate agent uses MCP tools vs manual operations | | verify_extension | Verify AGS MCP server in IDE extensions | | message_board | Message board operations (placeholder) | | precommit | Pre-commit MCP health checks | | smoke | MCP server smoke testing | | adapt | MCP adapter task wrapper |

Inputs

input.json fields:

Common:

  • operation (string, required): One of the operations listed above

For validate_access:

  • agent_action (string): Description of agent action
  • agent_code_snippet (string): Code the agent wrote
  • files_accessed (array): Files accessed manually
  • databases_queried (array): Databases queried directly

For verify_extension:

  • client (string): vscode, claude, or generic
  • entrypoint_substring (string): Path to MCP entrypoint
  • args (array): Arguments for entrypoint

For precommit:

  • entrypoint (string): Path to MCP entrypoint
  • auto_entrypoint (string): Path to auto entrypoint
  • args (array): Args for entrypoint
  • require_running (bool): Require running server PID
  • require_autostart (bool): Require Windows autostart
  • dry_run (bool): Skip execution

For smoke:

  • entrypoint_substring (string): Path to MCP entrypoint
  • args (array): Arguments for entrypoint
  • bridge_smoke (object): Bridge smoke test config

For adapt:

  • task (object): Task information with id field

For build:

  • task (object): Task information

For message_board:

  • (placeholder - not yet implemented)

Outputs

Output varies by operation. See individual skill documentation for details.

Constraints

  • Uses CORTEX query for file discovery (no direct filesystem scan)
  • Must not modify canon or context
  • Writes only under allowed roots via GuardedWriter
  • On non-Windows, running/autostart checks use PowerShell bridge

Fixtures

  • fixtures/basic/ - Basic operation tests

required_canon_version: >=3.0.0