XFMS MCP Server

XFMS — Xpansion Framework Model Source

Pick the right LLM for your task — without the Twitter vibes.

State what you're using the model for. XFMS aggregates evidence from eight independent benchmark sources, normalizes it onto a common scale, lets your intent decide which dimensions matter, and returns a ranked shortlist with plain-English rationale for every pick.

XFMS is one module of the Xpansion Framework — a unified architecture for governing AI-assisted work.

---

What this repository is

A thin Python client and command-line tool for calling the hosted XFMS API at xfms.vercel.app. About 250 lines of code. It turns a one-liner into a ranked LLM shortlist.

What this repository is not: the recommender engine, the score catalog, or the ingestion pipeline. Those run on the hosted service. The methodology behind every pick is published in full at docs/methodology.md — every claim there maps to code that runs at request time, you just don't run it locally.

---

What you say:

"Fixing bugs in our Python codebase."

What you get:

Top picks:
   1. 0.842  GPT-5.5                 (openai/gpt-5.5)         via OpenAI
   2. 0.811  Claude Opus 4.7         (anthropic/claude-opus-4.7) via Anthropic
   3. 0.798  Gemini 3.1 Pro Preview  (google/gemini-3.1-pro-preview) via Google

Inferred quality weights from your purpose:
  • structured_output_reliability  42.0%  ← BigCodeBench, Aider Polyglot
  • instruction_following          28.0%  ← LiveBench, Tau-Bench
  • factuality                     20.0%  ← MMLU, GPQA
  • coherence                      10.0%  ← LongBench

─── Explanation ───
Picked GPT-5.5: strong on structured output and instruction following —
the two dimensions that dominate code-edit work. Beats Claude on Aider
Polyglot and matches it on LiveBench reasoning, at roughly 60% of the
per-token cost.

Want to see how the picks actually behave on your kind of query? Add --ab:

─── A/B probe ───
Ran 5 test queries against the top picks.
  • GPT-4o-mini  avg_latency=5579 ms  total_cost=$0.00156  successes=5
  • GPT-5.5      avg_latency=8190 ms  total_cost=$0.07640  successes=5
  • GPT-5.4      avg_latency=8783 ms  total_cost=$0.03493  successes=5

Commentary:
  Across 5 real test queries, GPT-4o-mini was both cheapest ($0.0016 total)
  and fastest (5579 ms avg). Clear winner — 98% cheaper and 36% faster
  than the slowest pick.

---

What XFMS does for you

Beyond ranking, XFMS gives you these levers to honor what you actually meant:

• --primary <branch> — sacrosanct user preference. When you say "cheapest model, period", the engine switches to lexicographic ranking: cost wins, other dimensions only break ties. No more weighted-blend surprises.
• --ab — runs the top 3 picks against 5 real test queries (expanding to 10 or 15 if results trade off) and surfaces commentary on who won what. Grounds the recommendation in actual model behavior, not just benchmarks.
• --strict-priorities — when you name two co-equal drivers ("cheap but high quality too"), the engine refuses to silently blend; it asks you which way to break the tie.
• Latent-requirement suggestions — engine surfaces capabilities you didn't ask for but probably need (streaming for real-time chat, vision for OCR), so you don't get burned by what you didn't know.
• Deterministic by design — every internal model call is content- cached; same input always returns the same answer. The "I got different picks for the same question" failure mode is gone.

---

Install

pip install xfms

You need two free keys:

• Xpansion Framework Model Source access key — identifies you to the hosted API. Request one by submitting your email to the signup endpoint:

  curl -X POST https://xfms.vercel.app/signup \
    -H "Content-Type: application/json" \
    -d '{"email":"you@yourdomain.com"}'
  

  You'll get a confirmation email; click the button inside and your API key arrives in a follow-up email.

That's it — the hosted endpoint covers inference. There's no OpenRouter key to provision, no per-pick cost on you, no second account to manage. Configure the one key:

export XFMS_API_KEY=xfms_live_...

(Power users CAN supply their own OpenRouter key via OPENROUTER_API_KEY or --openrouter-key to route inference through their own account — but it's purely optional.)

Use

Command line:

xfms rank "writing a tight editorial under a budget"

xfms pick "fixing bugs in our Python codebase"

xfms rank "summarizing a long legal contract" --top-n 3

xfms rank "OCR a handwritten manifest" -c vision -c tool_use

When you actually mean "the cheapest model, period" — make cost the primary dimension and the engine switches to lexicographic ordering. The cheapest model wins; other dimensions only break ties:

xfms rank "cheapest model that can parse a 5-page PDF" --primary cost

When you want to see how the picks actually behave on your kind of query — add --ab and the engine runs the top 3 picks against 5 generated test queries (expanding to 10 or 15 if the picks trade wins), then surfaces real-world cost/latency plus plain-English commentary:

xfms rank "summarizing 50-page commercial leases" --ab

The A/B output ends with a one-paragraph summary along the lines of "On the test queries, Model X was 60% cheaper but Model Y was 30% faster — they trade off." You decide.

If XFMS detects something you didn't ask for but probably need — like streaming for a real-time chat use case — it surfaces a latent- requirement suggestion at the top of the response. The Koinonos lesson: sometimes you don't know what you don't know. Accept and re-run with -c, or ignore and ship.

Python:

from xfms_client import XFMSClient

with XFMSClient() as xfms:
    result = xfms.rank("writing a tight editorial under a budget")
    print(result["models"][0]["name"])

Or the one-shot:

from xfms_client import pick
print(pick("fixing bugs in our Python codebase")["name"])

---

Use it inside Claude Code, Cursor, or any MCP client

XFMS speaks Model Context Protocol (MCP) — the standard your AI assistant uses to call external tools. Once connected, you can ask the assistant "which model should I use for OCR on shipping manifests?" and it calls XFMS for you. No leaving the chat. No copy-pasting between windows.

Hosted install — one line, no install required

The XFMS engine hosts the MCP server itself at https://xfms.vercel.app/mcp/. No pip install, no OpenRouter key — just point your MCP client at the URL:

Claude Code:

claude mcp add xfms --transport http https://xfms.vercel.app/mcp/ \
  --header "Authorization: Bearer xfms_live_your_key_here"

Cursor (~/.cursor/mcp.json) — or paste through Settings → MCP:

{
  "mcpServers": {
    "xfms": {
      "url": "https://xfms.vercel.app/mcp/",
      "headers": {
        "Authorization": "Bearer xfms_live_your_key_here"
      }
    }
  }
}

Continue / Cline / any other MCP host — same URL + bearer header pattern; check your host's docs for the JSON config shape.

You need one key — the free XFMS access token. Request it at xpansion.dev/xfms/get-started or via curl; it arrives by email after you click the confirmation link. The hosted endpoint pays for the small inference call XFMS makes internally — when your host supports MCP sampling (Claude Code does), the call routes through your host's LLM and we don't pay either. Either way, you don't.

Restart your client, then ask it:

"Use XFMS to pick a model for summarizing long legal contracts."

Offline install — pip-installed local MCP server

For air-gapped environments or local development, the xfms Python package also ships a stdio MCP server you can install locally:

pip install 'xfms[mcp]'

Then register it with your host (uses xfms-mcp as the command):

claude mcp add xfms -- xfms-mcp \
  --env XFMS_API_KEY=xfms_live_your_key_here

The local server hits the same hosted endpoint, so the inference-cost behavior is identical to the hosted MCP install above.

Three tools are available to the assistant: rank (a ranked shortlist), pick (the single best pick), and discover (which quality dimensions matter for your purpose, without ranking).

One-click install via Smithery — the Smithery registry hosts a copy of this config so you can install without hand-editing JSON. Listed shortly after launch.

---

Override the system's inference

If you know which quality dimension matters most for your task, say so — your preference always wins over the LLM's inference:

xfms rank "code refactor" --leaf-priorities "structured_output_reliability=1.0,factuality=0.5"

xfms.rank(
    "code refactor",
    leaf_priorities={"structured_output_reliability": 1.0, "factuality": 0.5},
)

---

How XFMS picks — the four principles

Methodology in full at docs/methodology.md. The short version:

1. No provider self-reports. Every score comes from a third-party evaluator running the same protocol across every model.
2. No single-source dependence. Eight independent benchmark sources contribute today; no single leaderboard determines a pick.
3. User intent beats LLM inference. The system infers weights from your purpose, but your stated leaf_priorities always override the inference.
4. Honest gaps over invented signal. Missing data is recorded as missing — no interpolation, no synthetic scores. Coverage gaps surface on every pick.

---

Part of the Xpansion Framework

XFMS doesn't stand alone — it's the model-selection layer of the Xpansion Framework.

The Xpansion thesis

Humans communicate with intent compressed by contextual experience. AI simply predicts patterns in language. Xpansion is the execution layer that bridges them.

Every sentence a human types carries lifetimes of context that the speaker assumes the other side will decompress — what counts as "good enough," which constraints are non-negotiable, what failures last month taught them, what their house style demands. AI doesn't share that context. It predicts patterns in language, filling in the gaps with whatever's plausible to its training data. The result reads as plausible but isn't intent-honoring: sessions lose context, security holes ship silently, contracts break without warning, and there's no way to verify that what was built actually matches what was asked for. They don't know what they don't know, and neither does AI.

Xpansion closes the gap. It decompresses finite intent upfront, enforces it through code-driven AI behavior, and delivers binary-verified results against the intent across persistent memory that survives every session boundary.

Model Source — the model-selection enforcement

When you say "the best model for this task", you're compressing a lot: what counts as best depends on whether you care about factual reasoning or coherent prose, whether cost matters more than latency, whether you actually need vision or just text, whether the call has to stream, whether a particular benchmark dominates your real workload. AI on its own predicts the pattern — what model do most people pick for queries that look like this? — and gives you a plausible-sounding answer that's often wrong for you.

XFMS does the decompression. It takes your stated purpose, infers which benchmarks actually map to it, honors your stated primary preferences without silently overriding them, surfaces the latent requirements you didn't know to ask about (streaming for real-time chat, vision for OCR), and probes the top picks against your real query to verify the recommendation — not predict it. Then it tells you, in plain English, why it picked what it picked.

One module per enforcement

The rest of the Xpansion stack enforces the same decompress- enforce-verify contract for different parts of the work:

• Dispatch (Dispatch) — runtime task router. Watches what kind of work you're doing and routes it to the right tool.
• Finite Intent (XFFI) — turns "build me a feature" into a finite spec with binary terminals before any code gets written. Stops scope drift at the source.
• Boundary Auditor (XFBA) — checks every code edit against contracts. Stops broken function signatures and mismatched types from ever reaching production.
• Systemic Impact Analysis (XSIA) — maps the blast radius of a proposed change before it lands. Tells you what else might break.
• Token Conservation (XFTC) — manages how much of the conversation has to stay in the assistant's working memory. Prevents context loss in long sessions.
• Execution Audit (XFXA) — verifies every promise from the spec was actually met before declaring a task done. The final binary check.
• Memory Tree (XFMT) — session snapshots that stay searchable across conversations. Your assistant remembers what you decided last week.
• Security Auditor (XFSA) — static + AI security scanning on every code edit. Catches secrets, injection paths, and unsafe patterns before they ship.

The full picture, with the rest of the modules, lives at xpansion.dev.

Xpansion is in pre-signup right now. Early access and founding licenses are open at xpansion.dev. XFMS is the first piece to ship public + free — the rest follow.

---

Local development

git clone https://github.com/VisionAIrySE/XFMS.git
cd XFMS
python3 -m venv .venv
.venv/bin/pip install -e .[dev]
.venv/bin/python -m pytest tests/ -v

The tests mock the HTTP layer so they run offline — no API keys needed to develop.

---

License

This client library is MIT-licensed. The recommender engine, the catalog, and the ingestion pipeline are not open source. See NOTICE for the patent reservation language and the relationship to the broader Xpansion Framework IP.

---

Contact

• Russ Wright — russ@visionairy.biz
• Xpansion Framework — xpansion.dev
• Security disclosures — see SECURITY.md