Jcodeharness Mcp
Wire a remote Codex harness into your agent so code runs return structured verdicts, receipts, and usage logs instead of opaque terminal output.
Overview
JCode Harness Cloud is a MCP server for the Ship phase that connects agents to a paid remote OpenAI Codex harness and returns verdicts, receipts, and usage logs.
What is this MCP server?
- Paid remote streamable-http MCP at jcodeharness.clauxel.com with Bearer token auth
- Returns verdicts, receipts, usage logs, and audit-friendly structured output for Codex harness runs
- Targets OpenAI Codex harness MCP integrations for agent-driven coding workflows
- Publisher server card and GitHub repo: clauxel/jcode-harness-cloud-mcp
- Registry version 1.0.0 with agent-tools and paid-mcp tagging
- Single streamable-http remote endpoint with Authorization Bearer header
- Open-source repo: github.com/clauxel/jcode-harness-cloud-mcp
What problem does it solve?
Agent code runs often leave you guessing whether the harness actually passed, what it cost, and what to show a reviewer later.
Who is it for?
Solo builders using Codex or MCP agents who need receipt-backed harness results without self-hosting the harness stack.
Skip if: Teams that want a free local-only test runner with no Bearer token or paid remote dependency.
What do I get? / Deliverables
After install, harness invocations through MCP yield structured verdicts, receipts, and usage logs you can cite in ship and review workflows.
- Structured harness verdicts and receipts from remote tool calls
- Usage logs suitable for reviewing agent coding sessions
- JSON-oriented audit artifacts tied to Codex harness runs
Recommended MCP Servers
Journey fit
Harness verdicts and receipts sit in the Ship phase because they gate whether agent-generated code is acceptable before you merge or release. Testing is the canonical shelf: the server is built around execution outcomes and pass/fail-style verdicts from an OpenAI Codex harness workflow.
How it compares
Remote Codex harness MCP integration, not a local agent skill or generic unit-test framework.
Common Questions / FAQ
Who is JCode Harness Cloud for?
Indie and solo builders who run OpenAI Codex harness workflows through Claude Code, Cursor, or Codex and want verdicts and receipts over MCP.
When should I use JCode Harness Cloud?
Use it during Ship when you need structured harness outcomes, usage logs, and audit-friendly JSON before merging or releasing agent-generated code.
How do I add JCode Harness Cloud to my agent?
Add the streamable-http remote URL https://jcodeharness.clauxel.com/mcp in your MCP client and set Authorization to a Bearer token from the JCode Harness product website.