Toolkit
Capability 06 · reviewed 2026-08-01
On this page2 sections
MCP Integration
Model Context Protocol gives an AgentA worker you delegate to: brief it, and it takes steps on its own. A chatbot answers; an agent acts.An LLM that runs tools in a loop toward a goal — it acts, checks the result, decides the next step, and repeats until done.Full definition a negotiated interface to external tools and context. Claude Code, Codex, and Copilot all expose MCPUSB-C for AI — one standard plug so any agent can connect to any data source or tool.The Model Context Protocol: an open standard (Anthropic, 2024) for connecting AI systems to external tools and data without building a custom integration for each one.Full definition integrations, but the word “supports” hides material differences by surface.
Compare Primitives, Not Logos
For each target surface, verify:
- Which MCP primitives are consumed: tools, resources, prompts, or a subset.
- Which transports and authentication flows are supported.
- Where configuration and secrets live.
- Whether tool approval, allowlists, and managed policy are available.
- How connection failures, timeouts, and oversized responses are surfaced.
- Whether the integration runs locally, in a hosted environment, or both.
GitHub's cloud-agent documentation is a useful warning against family-level claims: that surface supports MCP tools while excluding resources, prompts, and remote OAuth. A statement about Copilot CLI MCP therefore cannot be copied to Copilot cloud agent unchanged.
Architecture Boundary
MCP is an integration protocol, not a trust boundary. A server's tool description is not authorization, and a successful connection is not evidence that its actions are safe. Apply least privilege at the credential and service layer, validate tool inputs, constrain egress, and record the resulting external action independently.
Prefer an existing official or reviewed server when its trust and output contracts fit. Build a custom server when domain semantics, access control, or response shaping are part of the system's correctness.
The deeper lenses retain Claude Code-specific configuration and server examples. Revalidate transport, authentication, and supported primitives for the exact product surface before reuse.
Documented coverage · 2026-08-01
MCP Integration by product
These classifications are documentation findings, not empirical runtime tests. Surface names are explicit because product-family claims are too coarse.
Claude Code
Anthropic
OpenAI Codex
OpenAI
Codex documents first-party MCP client configuration for extending local agent tool and context access.
Basis: documentedSources: 2
The reviewed MCP source is insufficient to assert one hosted capability, transport, and authentication contract.
Basis: documentedSources: 2
GitHub Copilot
GitHub
Copilot CLI provides first-party MCP server configuration and management.
Copilot cloud agent supports MCP tools, but the documented surface excludes resources, prompts, and remote OAuth.
Basis: documentedSources: 5
Source register
MCP Integration sources
Primary vendor documentation reviewed on the date shown. Links may change after publication.
- Connect Claude Code to tools via MCP (opens in a new tab)
Anthropic · official docs · accessed 2026-08-01
- Model Context Protocol (opens in a new tab)
OpenAI · official docs · accessed 2026-08-01
- Overview of customizing GitHub Copilot CLI (opens in a new tab)
GitHub · official docs · accessed 2026-08-01
- GitHub Copilot CLI command reference (opens in a new tab)
GitHub · official docs · accessed 2026-08-01
- MCP and Copilot cloud agent (opens in a new tab)
GitHub · official docs · accessed 2026-08-01
Comparative implementation depth