# specloop

> Human-expert guidance for AI coding agents. Send a coding question with your app context; a vetted human expert answers; you get an agent-ready prompt back. Interfaces: a CLI (`@specloop/cli`) and an MCP server (local stdio + remote HTTP).

## Docs
- [Full documentation (Markdown)](https://specloop.dev/docs.md): Quickstart, CLI install, credits, MCP setup for Claude Code / Claude Desktop / Codex / Cursor, remote MCP, tools, command reference, experts.
- [Docs (web)](https://specloop.dev/docs): human-readable version of the same content.

## Quick setup
- CLI: `npm i -g @specloop/cli` then `specloop login`
- MCP (stdio, after `specloop login`): command `npx`, args `["-y","@specloop/cli","mcp"]`
- Remote MCP: `https://specloop.dev/mcp` (Streamable HTTP, header `Authorization: Bearer <token>`)

## MCP tools
- submit_human_guidance(query, summary, techStack, tools, codeSnippet?, githubUrl?)
- get_human_guidance_status(requestId)
- get_expert_comments(requestId)
- get_agent_ready_prompt(requestId)