بنقرة واحدة
mcp-integration
How to configure, connect to, and use MCP (Model Context Protocol) servers in go-orca workflows.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
القائمة
How to configure, connect to, and use MCP (Model Context Protocol) servers in go-orca workflows.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Offload expensive agent work to a go-orca cluster via the mcp-orca MCP bridge (workflows and ad-hoc persona runs).
Use this skill when implementing code for any module in this repository.
Use when implementing or reviewing Next.js App Router applications in go-orca workflows.
Model routing and structured-output rules for go-orca workflow personas (Director, Project Manager, Architect).
Techniques for writing clear, compelling technical blog posts and documentation.
Checklists and patterns for systematic QA review of code, APIs, and infrastructure changes.
| name | mcp-integration |
| description | How to configure, connect to, and use MCP (Model Context Protocol) servers in go-orca workflows. |
Use this skill when configuring tool access via MCP servers or when an agent needs to interact with external tools exposed over MCP.
The Model Context Protocol (MCP, 2025 spec) is a JSON-RPC based protocol that lets AI models discover and call external tools through a standardized interface. go-orca supports three transport types:
| Transport | When to use |
|---|---|
streamable (default) | Modern HTTP-based MCP servers (2025-03-26 spec) |
sse | Legacy HTTP servers using Server-Sent Events (2024-11-05 spec) |
command | Local subprocess servers run over stdio |
Add an MCP source to the tools.mcp section of your go-orca.yaml:
tools:
mcp:
- name: fetch
endpoint: "http://localhost:3000/mcp"
transport: streamable # or "sse"
- name: local-tools
command: uvx
args: ["mcp-server-fetch"]
transport: command
See mcp-config.yaml for a complete annotated example.
On startup go-orca calls ListTools on each configured MCP server and registers the discovered tools in the global tool registry. Agents can then invoke these tools by name in their tools list.
Use the check-mcp-server.sh script to verify that an MCP server is reachable and returns a valid tool list before adding it to configuration.
See mcp-server-catalog.md for a curated list of open-source MCP servers and their endpoint conventions.
isError: true, go-orca surfaces the error message from the first text content block.Load are fatal for that MCP source; the workflow continues with tools from other sources.tls_skip_verify: true only in local/dev environments — never in production.When tools.toolchains maps a workflow to mcp-nextjs-toolchain or mcp-node-toolchain, the engine runs workspace preflight before install/build:
| Check | Example failure |
|---|---|
Strict JSON package.json | Leading // comment lines |
| Real build script | "build": "echo build successful" |
| PostCSS deps | postcss.config.js references tailwindcss but package.json lacks it |
| Route conflicts | Both app/page.js and app/page.tsx |
| Router conflict | Both app/page.* and pages/index.* |
MCP resources:
orca://schemas/package.json — manifest formatorca://schemas/nextjs-preflight — full Next.js preflight checklistBuild capabilities (npm_build, next_build) also reject fake scripts at the MCP server layer.