syfert-mcp
Manifests and install instructions for the Syfert legal-research MCP server - verify every citation, check whether a case is still good law, audit a whole brief. 10.7M U.S. opinions + statutes for all 50 states.
Install / Use
claude mcp add Gwonk1 -- npx -y github:Gwonk1/syfert-mcpIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
Education & ResearchSupported Platforms
Tags
Our assessment of syfert-mcp
syfert-mcp scores 81/100 on our quality scale, 332nd of 400 Education & Research skills we index.
Its MCP Server is 11 KB long, well organised into 20 sections with 13 code examples: a thorough specification that gives an agent plenty to work with.
It has 3 GitHub stars, so there is little community track record yet; judge it on its content.
Maintenance, license and trust
- The repository was last updated 21 days ago, so syfert-mcp is actively maintained.
- No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
- Its trust signals score 80/100, with 2 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands.
Automated pattern scan on 2026-10-01. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
syfert-mcp compared with similar skills
All 4 of these similar skills score higher than syfert-mcp; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| syfert-mcp (this skill)by Gwonk1 | 81 | 3 | 21d ago | MCP Server |
| Agent-Reachby Panniantong | 100 | 87.2k | 16d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 74.2k | today | CLAUDE.md |
| rufloby ruvnet | 100 | 73.6k | today | CLAUDE.md |
| last30days-skillby mvanhorn | 100 | 63.3k | today | CLAUDE.md |
Frequently asked questions
- How do I install syfert-mcp?
- Run
claude mcp add Gwonk1 -- npx -y github:Gwonk1/syfert-mcp. The install tabs above show the steps for each supported agent. - Which AI agents does syfert-mcp work with?
- It is written for Claude Code, Claude Desktop and Gemini CLI, as a MCP Server file. Other agents that read the same format can often use it too.
- Is syfert-mcp safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. It declares no license and scores 80/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is syfert-mcp still maintained?
- The repository was last updated 21 days ago, so syfert-mcp is actively maintained.
Skill content
View source on GitHubSyfert Legal Research — MCP server
The verification layer for legal AI output. Verify every citation, check whether a case is still good law, audit a whole brief in one pass — and search statutes for every state.
- Endpoint:
https://mcp.syfert.com/mcp(Streamable HTTP) - Auth:
Authorization: Bearer <token>— optional, see below - Free token: one email click at https://syfert.com/mcp/ — no card, no trial clock
- Docs: https://syfert.com/mcp/ · Tool detail: https://syfert.com/mcp/tools.php
- Maintainer: Graham W. Syfert, Esq., P.A. — graham@syfert.com
No token? Just connect.
Point any MCP client at https://mcp.syfert.com/mcp with no credential at all and it answers.
All 13 tools are read-only, so anonymous access is the whole product, not a demo of it —
it is simply metered per IP address: 20 calls per minute and 200 calls per day.
initialize and tools/list are never counted, so discovery always works.
A free token — one email click at https://syfert.com/mcp/ — lifts those limits to 120 calls
per minute, restores the full result caps and the 512 KB check_brief document, and attaches
your calls to your own account. Send it as Authorization: Bearer YOUR_TOKEN.
curl -s https://mcp.syfert.com/mcp -H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
What is behind it
| | | |---|---| | Case law | 10.7 million U.S. opinions — every state court and every federal court (Free Law Project / CourtListener bulk data, refreshed quarterly) | | Citation map | ~175 million citation events — each place one opinion cites another, with the judge's sentence, parenthetical, quoted words and pin cite | | Treatment | Syfertize citator flags: red = overruled or superseded, yellow = questioned or limited, green = followed | | Statutes | All 50 states + D.C., plus the U.S. Code, the C.F.R., the federal rules and the sentencing guidelines — 55 corpora, each with its own index and scrape date | | Quotes | 448,000 passages that courts quote from each other, each tied to its source opinion |
All 13 tools are read-only. Nothing writes, nothing bills, nothing leaves the corpus.
Tools
| Tool | What it does |
|---|---|
| search_cases | Full-text search across 10.7M opinions — boolean, phrases, proximity, court/date/state filters, authority-weighted ranking. |
| get_case | Full opinion text (majority, concurrence, dissent) with metadata, parallel citations, the Bluebook cite and the treatment flag. Long opinions page. |
| check_citation | The hallucination catcher: does the cite exist, what case really lives there, does the claimed case name match. Misses get did-you-mean suggestions. |
| check_brief | Whole-document audit — every citation, every claimed case name and every quoted passage in a finished draft, verified in one pass. |
| get_treatment | Is it still good law? Red/yellow/green with the citing cases behind each signal, who overruled it (including statutory supersession), and still-good replacement authority. |
| get_propositions | What courts actually cite the case for, proposition by proposition, each with its own flag — plus a quote-integrity verdict per famous passage. |
| get_citing_cases | The citator list: ranked citing cases with judge-written parentheticals, verbatim quotes, pin cites and each citing court's own signal. |
| find_related_cases | Hand it one good case, get the doctrinal line it belongs to — the frequently co-cited family around that anchor authority. |
| search_quotes | Phrase search across 448,000 quoted passages, ranked by adoption, each tied to the opinion it originates in. Catches misattributed quotes. |
| find_issues | Start from facts or an issue instead of a case name: the doctrine as courts in that state name it, the cases on each side, the statutes those opinions construe. |
| find_court | Turns "the Eleventh Circuit" or "S.D. Florida" into the court identifier the search filters expect. |
| search_statutes | Search statutes and codes for all 50 states + D.C., the U.S. Code, the C.F.R., the federal rules and the sentencing guidelines. |
| get_statute | Full section text with cross-references surfaced and Notes-of-Decisions annotations; federal sections carry a banner when they have been held unconstitutional. |
Install
Every snippet below sends a token. You do not have to — drop the Authorization header (or
use the plain endpoint URL) and the same client connects anonymously at 20 calls/min, 200/day.
To lift that: https://syfert.com/mcp/ → enter your email → open the one-time link, then
replace YOUR_TOKEN everywhere below with what it gives you. Treat it like a password.
Claude Code
claude mcp add --transport http syfert https://mcp.syfert.com/mcp \
--header "Authorization: Bearer YOUR_TOKEN"
claude.ai (custom connector)
claude.ai custom connectors cannot set request headers, so they use the legacy path form. Prefilled link — paste your own token into the URL first:
https://claude.ai/customize/connectors?modal=add-custom-connector&connectorName=Syfert&connectorUrl=https%3A%2F%2Fmcp.syfert.com%2FYOUR_TOKEN%2Fmcp
Or by hand: Settings → Connectors → Add custom connector → https://mcp.syfert.com/YOUR_TOKEN/mcp.
ChatGPT (developer mode)
Settings → Apps & Connectors → Advanced → Developer mode, then Create.
URL https://mcp.syfert.com/YOUR_TOKEN/mcp, authentication No authentication
(the token is already in the URL). ChatGPT cannot set headers either.
Codex CLI
export SYFERT_MCP_TOKEN=YOUR_TOKEN
codex mcp add syfert --url https://mcp.syfert.com/mcp \
--bearer-token-env-var SYFERT_MCP_TOKEN
Cursor
One-click — paste your token in first (see "Building your own deeplink" below):
cursor://anysphere.cursor-deeplink/mcp/install?name=syfert&config=eyJ1cmwiOiJodHRwczovL21jcC5zeWZlcnQuY29tL21jcCIsImhlYWRlcnMiOnsiQXV0aG9yaXphdGlvbiI6IkJlYXJlciBZT1VSX1RPS0VOIn19
Or edit ~/.cursor/mcp.json (see mcp.json in this repo).
The config parameter is the base64 of this object with your token in it:
{"url":"https://mcp.syfert.com/mcp","headers":{"Authorization":"Bearer YOUR_TOKEN"}}
python3 - <<'PY'
import json, base64, urllib.parse
cfg = {"url": "https://mcp.syfert.com/mcp",
"headers": {"Authorization": "Bearer YOUR_TOKEN"}}
b = base64.b64encode(json.dumps(cfg, separators=(',', ':')).encode()).decode()
print("cursor://anysphere.cursor-deeplink/mcp/install?name=syfert&config=" +
urllib.parse.quote(b, safe=''))
PY
</details>
VS Code / GitHub Copilot
This link prompts for the token and stores it as a secret input:
vscode:mcp/install?%7B%22name%22%3A%22syfert%22%2C%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fmcp.syfert.com%2Fmcp%22%2C%22headers%22%3A%7B%22Authorization%22%3A%22Bearer%20%24%7Binput%3Asyfert_token%7D%22%7D%2C%22inputs%22%3A%5B%7B%22type%22%3A%22promptString%22%2C%22id%22%3A%22syfert_token%22%2C%22description%22%3A%22Syfert%20MCP%20token%22%2C%22password%22%3Atrue%7D%5D%7D
Or from the command line: code --add-mcp '{"name":"syfert","type":"http","url":"https://mcp.syfert.com/mcp","headers":{"Authorization":"Bearer YOUR_TOKEN"}}'
Gemini CLI
As an extension (this repo — it carries gemini-extension.json):
gemini extensions install https://github.com/Gwonk1/syfert-mcp
You will be asked for the Syfert MCP Token setting on install; it is stored in your system keychain, not in the manifest. Or add the server directly:
gemini mcp add syfert https://mcp.syfert.com/mcp --transport http \
-H "Authorization: Bearer YOUR_TOKEN"
Goose
One-click (Goose will prompt for the Authorization header value):
goose://extension?type=streamable_http&url=https%3A%2F%2Fmcp.syfert.com%2Fmcp&id=syfert&name=Syfert%20Legal%20Research&description=Verify%20citations%2C%20check%20whether%20a%20case%20is%20still%20good%20law%2C%20audit%20a%20whole%20brief%2C%20and%20search%20statutes%20for%20all%2050%20states&header=Authorization%3DBearer%20%3CYOUR_SYFERT_TOKEN%3E
Or: Extensions → Add custom extension → type Streamable HTTP, URL
https://mcp.syfert.com/mcp, request header Authorization: Bearer YOUR_TOKEN.
Windsurf
~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"syfert": {
"serverUrl": "https://mcp.syfert.com/mcp",
"headers": { "Authorization": "Bearer YOUR_TOKEN" }
}
}
}
Any client that only speaks stdio (mcp-remote fallback)
{
"mcpServers": {
"syfert": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://mcp.syfert.com/mcp",
"--header", "Authorization: Bearer YOUR_TOKEN"]
}
}
}
Two URL forms
| Form | Use it when |
|---|---|
| https://mcp.syfert.com/mcp + Authorization: Bearer <token> | Preferred. Your client can set request headers (everything above except claude.ai and ChatGPT). |
| https://mcp.syfert.com/<token>/mcp | Legacy fallback for clients that cannot set headers — claude.ai custom connectors, ChatGPT developer mode. |
Both reach the same server. An unauthenticated request returns 401 with a
WWW-Authenticate: Bearer challenge and RFC 9728 protected-resource metadata.
Privacy — what is and is not logged
Your searches are not logged. Not the citation you checked, not the case name, not your query, not a brief you pasted in. What is kept is operational only: which tool ran, whether it succeeded, and how long it took — enough to keep the lights on and catch abuse. That data is seen by Graham W. Syfert, Esq., P.A. and nobody else. It is not sold, not shared with third parties, and not fed to any advertising system.
Your token is a secret; anyone holding it can search against your access. If it leaks, email graham@syfert.com and a fresh one will be issued and the old one retired.
Be sensible anyway: do not paste client secrets into a search box, here or anywhere.
- Privacy policy: https://syfert.com/privacy
- Terms: https://syfert.com/terms
Support
Email graham@syfert.com, or open an issue on this repo. Bugs get fixed.
What is in this repo, and what is not
This repository holds manifests and install instructions only — the JSON files that let MCP clients and catalogs find the hosted service. It is published under the MIT licence so catalogs can vendor and redistribute the manifests freely.
The Syfert service itself is proprietary and closed-source. The corpus builders, the Syfertize treatment engine, the citation-event pipeline and the server code are not published here or anywhere. Nothing in this repository grants any right to the service, its data, or its output beyond the terms.
Not legal advice
Syfert is a research tool. It checks the citations; you are still the lawyer, and the filings are still yours. Using it creates no attorney-client relationship.
Related Skills
Agent-Reach
87.2kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
74.2kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
ruflo
73.6k🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
last30days-skill
63.3kAI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
