Consult7
Offload large-context file analysis to 500+ OpenRouter models (up to 2M tokens) from any MCP agent.
Add to your client
Copy the config for your MCP client and paste it into its config file.
claude mcp add -s user consult7 uvx -- consult7 your-openrouter-api-keyPaste into ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"consult7": {
"command": "uvx",
"args": [
"consult7",
"your-openrouter-api-key"
]
}
}
}Requires `uv` (the Python package runner). Install it from https://docs.astral.sh/uv/ if `uvx` is not found.
Step-by-step guides: Add to Claude Desktop · Add to Cursor · Add to Windsurf
Before you start
- An OpenRouter API key
- uvx (from uv) to download and run the server in an isolated environment
- An MCP-compatible client such as Claude Code or Claude Desktop
About Consult7
Consult7 is an MCP server that enables any MCP-compatible agent to offload file analysis to large context window models (up to 2M tokens) via OpenRouter. You provide absolute file paths (with optional filename wildcards) and a query; Consult7 assembles the files into a single context, sends them to a chosen model in a selected reasoning mode, and returns the result. It supports all 500+ OpenRouter models, dynamic per-model file size limits, an optional output file to avoid flooding context, and OpenRouter Fusion multi-model panels.
Tools & capabilities (1)
consultationCollects files from the given absolute paths (wildcards allowed in filenames only), assembles them into a single context, and sends them with your query to a chosen OpenRouter model. Parameters: files (required list of absolute paths/patterns), query (required), model (required OpenRouter model ID), mode (required: fast | mid | think), output_file (optional absolute path to save the response, returns only a confirmation message), and zdr (optional bool to enable Zero Data Retention routing, default false).
When to use it
- Quickly summarize the architecture and main components of a codebase in a single query
- Run a deep, step-by-step security analysis of an authentication flow with a reasoning model
- Generate a comprehensive code review report saved to a file instead of into the agent's context
- Analyze large file collections that exceed the calling agent's context window
- Compare answers from multiple frontier models, or use Fusion to have a panel deliberate and a judge synthesize one answer
Security notes
Requires an OpenRouter API key, passed as a command-line argument. Files matching __pycache__, .env, secrets.py, .DS_Store, .git, and node_modules are automatically ignored. File contents are sent to the selected OpenRouter model; an optional zdr parameter enables Zero Data Retention routing (available for Gemini 3.1 Pro/Flash, Claude Opus 4.8, GPT-5, GPT-5.5; not available for Grok 4.20).
Consult7 FAQ
Do I need to install anything?
No. uvx automatically downloads and runs consult7 in an isolated environment; you only need an OpenRouter API key.
Which models can I use?
All 500+ models available on OpenRouter via their model IDs (e.g. google/gemini-3.1-pro-preview, anthropic/claude-opus-4.8, x-ai/grok-4.20). File size limits are calculated automatically from each model's context window.
How do file paths work?
Use absolute paths only. Wildcards are allowed in filenames but not in directory paths, and an extension is required with wildcards (e.g. /Users/john/project/src/*.py). You can mix explicit files and patterns in one list.
What are the performance modes?
fast (no reasoning, quick answers), mid (moderate reasoning, code reviews and bug analysis), and think (maximum reasoning, security audits and complex refactoring). For Fusion, mode maps to web-research depth.
How do I avoid flooding my agent's context with a long response?
Pass output_file with an absolute path. The response is written to disk and the tool returns only a confirmation message; if the file exists it is saved with an _updated suffix.
Alternatives to Consult7
Compare all alternatives →
Official MCP server providing persistent, file-backed knowledge-graph memory across sessions.
Structured step-by-step reasoning tool for breaking problems into revisable thought sequences.
Fully managed remote server for AWS docs, blogs, What's-New and Well-Architected guidance — no key.
Compare Consult7 with: