QuReDec
Evidence-backed decision briefs with citations — recommendation, confidence, risks, sources.
Runs on your machine from its package; the steps are below.
What it does
- Reads
- Writes
- High impact
At most: only reads or looks up information.
What it costs
FreeOdel charges nothing per call. The publisher may charge for its own service.
Runs locally over stdio
This server isn't hosted — your MCP client launches it from a package registry. Use one of the commands below, or drop the config into your client (e.g. Claude Desktop).
Run
uvx quredec-mcp
MCP client config
{
"mcpServers": {
"quredec": {
"command": "uvx",
"args": [
"quredec-mcp"
]
}
}
}QuReDec MCP Server
Run a structured QuReDec decision brief from inside Claude Code, Claude Desktop, Cursor, or any MCP-compatible client.
Status: alpha (2026-04-30). Wired to the live QuReDec public API (
/api/v1/brief*, per-user Bearer-token auth). SetQUREDEC_MOCK=1for offline development.
What this does
| Tool | Description |
|---|---|
decision_brief | Submit a question; receive a structured, evidence-backed decision brief with citations |
get_brief_status | Poll an in-progress brief by ID |
list_recent_briefs | List the caller's recent briefs |
The completed brief contains: executive_summary, recommendation,
confidence (0–1), key_facts[].citation_ids, implications[], actions[]
(with effort + impact), risks[] (with severity + citation_ids),
and citations[] (with citation_id, title, url, source_type).
Install
pip install quredec-mcp
Or from source:
git clone https://github.com/Advanced-Binary-Operations/QuReDec_MCP
cd QuReDec_MCP
pip install -e .
Configure
Generate a QuReDec API key at https://quredec.com/account → API keys. Each key is shown to you exactly once at creation; copy it into a password manager immediately.
Add to your MCP client config (Claude Desktop example):
{
"mcpServers": {
"quredec": {
"command": "quredec-mcp",
"env": {
"QUREDEC_API_KEY": "qrd_live_xxxxxxxxxxxx"
}
}
}
}
Optional env vars:
QUREDEC_BASE_URL— defaults tohttps://quredec.com.QUREDEC_MOCK=1— run against an in-process mock for development.
Usage example
In Claude Code or Claude Desktop:
Use the
decision_brieftool to evaluate "Should I migrate from Stripe to LemonSqueezy?"
The brief comes back with: recommendation, confidence score, key facts with inline citation ids, risks, recommended actions, and a public share URL. Briefs typically take 2–5 minutes; the tool blocks and polls by default (timeout 600 s) and falls back to a synthetic running status with a poll hint if the timeout fires.
Pricing
The MCP server is free. Briefs consume credits from your QuReDec account:
- Starter $4.99 — 3 briefs (one-time)
- Pro $49/mo — 30 briefs/mo
See https://quredec.com/pricing.
Development
Unit tests run against the in-process mock (no network, no key required):
pip install -e ".[dev]"
QUREDEC_MOCK=1 pytest
Run the server locally against the mock:
QUREDEC_MOCK=1 quredec-mcp
Live integration test
scripts/integration_test.py exercises the full path against quredec.com:
submit → poll → assert brief schema → list. It consumes 1 credit from the
account whose key you pass in, so run it sparingly.
QUREDEC_API_KEY=qrd_live_xxxxxxxxxxxx \
python scripts/integration_test.py
Optional env vars:
QUREDEC_BASE_URL— defaults tohttps://quredec.com.QUREDEC_QUESTION— override the default seed question.QUREDEC_TIMEOUT— poll timeout in seconds (default 600).
Exit code 0 on full pass, 1 on the first failed assertion.
License
MIT — see LICENSE.
Maintainer
Advanced Binary Operations LLC · support@advancedbinaryoperations.com
