Prototype · Preview: every server in the official registry, synced every 30 minutes. Sign in with Google to write reviews.Prototype · updated every 30 min
PopNx MCPMCP servers for people who use AI · explained in plain words

Mendrift

by suneel190700version 0.1.2i★ 0iRuns on your computerRepository ↗
Registry nameio.github.suneel190700/mendrift-mcp

How to install

What it does

MCP server for ML drift detection, MLflow registry diffing, and HMAC-gated model rollback

  • Connects Runs on your computer
  • Account or key No key needed
  • Category Other

Why it ranks here

Lists are ordered by a usage score (out of 100) built from these parts

Popularity0/80GitHub stars or weekly downloads: the larger counts 60, the smaller 20
Freshness3/12Full marks for a release or code change within 30 days
Still answering0/8The cloud endpoint answered at the last check
Usage score3/100

Quality: Passed 3/3 basic checks (shown, not used for ordering) · Momentum: No data yet (Used only on the Trending tab)

Install

Pick the AI app you use, then copy the settings.

Let your AI install it with one sentence

Please install Mendrift. Follow https://mcp.popnx.com/i/suneel190700-mendrift-mcp.md, show me the commands and wait for my OK before running them.

Paste it into an AI that can work on your computer (such as Claude Code or Cursor). It reads our steps, shows you the commands and waits for your OK. No key, no sign-up. · See the full text your AI reads

Passed 3/3 checks · Runs on your computer · ✓ Commands checked Nov 14, 2026

First: which AI do you use?

Pick your app and the steps appear below.

Works with Claude, with some setup first

Install Python (uv) on your computer first (one time) · Only the Claude desktop app (Windows or Mac), not the web or phone

Let AI install it

Copy this message into a Claude chat. The AI checks your computer, tells you which file it will change, and waits for your OK every time.

Please install the MCP server Mendrift for Claude on this computer, in this order: 1) Check uv with uv --version. If it is missing, tell me to install it from https://docs.astral.sh/uv/getting-started/installation/; do not install it yourself. 2) Use only this config (already checked by PopNx MCP): { "mcpServers": { "mendrift": { "command": "uvx", "args": [ "mendrift-mcp" ] } } } 3) Tell me the full path of the file you will change, or the command you will run, and wait for my OK. 4) No key is needed. 5) Do what we agreed, then check that the Mendrift tools really show up.

This message has no key in it. If a key is needed, the AI will ask you to paste it yourself into the file or the app’s field.

  1. Install uv (if you don’t have it)

    Already installed? Check first · Open Terminal, type this command and press Enter

    uv --version
    • A version number such as uv … = installed, skip this step
    • "command not found" (on Windows: "is not recognized") = not installed yet, or installed but Terminal needs to be closed and reopened
    How to open Terminal

    Mac: press ⌘ + Space, type Terminal, press Enter

    Windows: open Start, type Windows PowerShell, then Open

    Not installed: uv has no click-through installer. Copy the one-line command for your system (macOS or Windows) from uv’s official page into Terminal · no separate Python install needed; uv downloads it when required

    Open the uv download page ↗

    When it’s done, close Terminal, open it again and run the same command. You should see a version number: uv --version

    Install once; it works for everything that needs uv

  2. Open the Claude desktop app’s settings file

    SettingsDeveloperEdit Config

  3. Paste these settings and save the file

    {
      "mcpServers": {
        "mendrift": {
          "command": "uvx",
          "args": [
            "mendrift-mcp"
          ]
        }
      }
    }
  4. Quit Claude and open it again

  5. Try it

    Use Mendrift and tell me what you can do for me, then show me one thing

    ✓ If Claude says it is using Mendrift or shows a Mendrift tool in the answer, you are connected

Menu names checked against Anthropic’s help page on Nov 11, 2026 · your screen may differ slightly by app version

ChatGPT cannot connect Mendrift yet

Mendrift runs on your computer, but ChatGPT can only connect servers that run on the web

What you can do instead

A · Use an app that can connect Mendrift

  • Install Python (uv) on your computer first (one time) · Only the Claude desktop app (Windows or Mac), not the web or phone
  • Install Python (uv) on your computer first (one time)
  • Install Python (uv) on your computer first (one time)
  • Install Python (uv) on your computer first (one time)

B · Others in Other that work with ChatGPT

Picked from the most used in the same category, only ones that connect by address with no key

Works with Claude Code, with some setup first

Install Python (uv) on your computer first (one time)

Let AI install it

Copy this message into a Claude Code chat. The AI checks your computer, tells you which file it will change, and waits for your OK every time.

Please install the MCP server Mendrift for Claude Code on this computer, in this order: 1) Check uv with uv --version. If it is missing, tell me to install it from https://docs.astral.sh/uv/getting-started/installation/; do not install it yourself. 2) Use only this config (already checked by PopNx MCP): claude mcp add --scope user mendrift -- uvx mendrift-mcp 3) Tell me the full path of the file you will change, or the command you will run, and wait for my OK. 4) No key is needed. 5) Do what we agreed, then check that the Mendrift tools really show up.

This message has no key in it. If a key is needed, the AI will ask you to paste it yourself into the file or the app’s field.

  1. Install uv (if you don’t have it)

    Already installed? Check first · Open Terminal, type this command and press Enter

    uv --version
    • A version number such as uv … = installed, skip this step
    • "command not found" (on Windows: "is not recognized") = not installed yet, or installed but Terminal needs to be closed and reopened
    How to open Terminal

    Mac: press ⌘ + Space, type Terminal, press Enter

    Windows: open Start, type Windows PowerShell, then Open

    Not installed: uv has no click-through installer. Copy the one-line command for your system (macOS or Windows) from uv’s official page into Terminal · no separate Python install needed; uv downloads it when required

    Open the uv download page ↗

    When it’s done, close Terminal, open it again and run the same command. You should see a version number: uv --version

    Install once; it works for everything that needs uv

  2. Open Terminal

    Mac: press Command+Space and type Terminal · Windows: open PowerShell

  3. Paste this command and press Enter

    claude mcp add --scope user mendrift -- uvx mendrift-mcp
  4. Type /mcp in Claude Code

    You should see Mendrift listed as connected

  5. Try it

    Use Mendrift and tell me what you can do for me, then show me one thing

    ✓ If Claude Code says it is using Mendrift or shows a Mendrift tool in the answer, you are connected

Menu names checked against Anthropic’s help page on Nov 11, 2026 · your screen may differ slightly by app version

Works with Cursor, with some setup first

Install Python (uv) on your computer first (one time)

Let AI install it

Copy this message into a Cursor chat. The AI checks your computer, tells you which file it will change, and waits for your OK every time.

Please install the MCP server Mendrift for Cursor on this computer, in this order: 1) Check uv with uv --version. If it is missing, tell me to install it from https://docs.astral.sh/uv/getting-started/installation/; do not install it yourself. 2) Use only this config (already checked by PopNx MCP): { "mcpServers": { "mendrift": { "command": "uvx", "args": [ "mendrift-mcp" ] } } } 3) Tell me the full path of the file you will change, or the command you will run, and wait for my OK. 4) No key is needed. 5) Do what we agreed, then check that the Mendrift tools really show up.

This message has no key in it. If a key is needed, the AI will ask you to paste it yourself into the file or the app’s field.

  1. Install uv (if you don’t have it)

    Already installed? Check first · Open Terminal, type this command and press Enter

    uv --version
    • A version number such as uv … = installed, skip this step
    • "command not found" (on Windows: "is not recognized") = not installed yet, or installed but Terminal needs to be closed and reopened
    How to open Terminal

    Mac: press ⌘ + Space, type Terminal, press Enter

    Windows: open Start, type Windows PowerShell, then Open

    Not installed: uv has no click-through installer. Copy the one-line command for your system (macOS or Windows) from uv’s official page into Terminal · no separate Python install needed; uv downloads it when required

    Open the uv download page ↗

    When it’s done, close Terminal, open it again and run the same command. You should see a version number: uv --version

    Install once; it works for everything that needs uv

  2. Click this button to open Cursor

    Open in Cursor

    Cursor asks you to confirm; click Install

  3. Open the AI chat in Cursor in Agent mode

  4. Try it

    Use Mendrift and tell me what you can do for me, then show me one thing

    ✓ If Cursor says it is using Mendrift or shows a Mendrift tool in the answer, you are connected

Menu names checked against Cursor’s help page on Nov 11, 2026 · your screen may differ slightly by app version

Works with VS Code, with some setup first

Install Python (uv) on your computer first (one time)

Let AI install it

Copy this message into a VS Code chat. The AI checks your computer, tells you which file it will change, and waits for your OK every time.

Please install the MCP server Mendrift for VS Code on this computer, in this order: 1) Check uv with uv --version. If it is missing, tell me to install it from https://docs.astral.sh/uv/getting-started/installation/; do not install it yourself. 2) Use only this config (already checked by PopNx MCP): { "servers": { "mendrift": { "type": "stdio", "command": "uvx", "args": [ "mendrift-mcp" ] } } } 3) Tell me the full path of the file you will change, or the command you will run, and wait for my OK. 4) No key is needed. 5) Do what we agreed, then check that the Mendrift tools really show up.

This message has no key in it. If a key is needed, the AI will ask you to paste it yourself into the file or the app’s field.

  1. Install uv (if you don’t have it)

    Already installed? Check first · Open Terminal, type this command and press Enter

    uv --version
    • A version number such as uv … = installed, skip this step
    • "command not found" (on Windows: "is not recognized") = not installed yet, or installed but Terminal needs to be closed and reopened
    How to open Terminal

    Mac: press ⌘ + Space, type Terminal, press Enter

    Windows: open Start, type Windows PowerShell, then Open

    Not installed: uv has no click-through installer. Copy the one-line command for your system (macOS or Windows) from uv’s official page into Terminal · no separate Python install needed; uv downloads it when required

    Open the uv download page ↗

    When it’s done, close Terminal, open it again and run the same command. You should see a version number: uv --version

    Install once; it works for everything that needs uv

  2. Click this button to open VS Code

    Open in VS Code

    VS Code asks you to confirm; click Install

  3. Open the AI chat in VS Code in Agent mode

  4. Try it

    Use Mendrift and tell me what you can do for me, then show me one thing

    ✓ If VS Code says it is using Mendrift or shows a Mendrift tool in the answer, you are connected

Menu names checked against VS Code’s help page on Nov 11, 2026 · your screen may differ slightly by app version

Other apps that use MCP, such as Windsurf, Zed or Gemini CLI, each add servers differently. See how in the app list.

See every app that uses MCP →

JSON settings are under “For developers” below

Stuck? Ask people who use Mendrift💬 Discussion →
For developers: Terminal commands and JSON settings

Installed on your computer

Claude Code

claude mcp add --scope user mendrift -- uvx mendrift-mcp

Claude Desktop (claude_desktop_config.json)

{
  "mcpServers": {
    "mendrift": {
      "command": "uvx",
      "args": [
        "mendrift-mcp"
      ]
    }
  }
}

Cursor (~/.cursor/mcp.json)

{
  "mcpServers": {
    "mendrift": {
      "command": "uvx",
      "args": [
        "mendrift-mcp"
      ]
    }
  }
}

VS Code (mcp.json)

{
  "servers": {
    "mendrift": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "mendrift-mcp"
      ]
    }
  }
}

These settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.

At a glance

Downloads per weeki
17/wk
Connection
Runs on your computer
Latest versioni
0.1.2
Categoryi
Other
Published to registryi
Jul 20, 2026
Code last changedi
Aug 9, 2026
Maker
suneel190700
Packages
pypi: mendrift-mcp
Registry status
Active

Which apps can use it

Based on how this server connects and on each app’s own docs, checked Nov 10, 2026.

Run on your computer in

ClaudeGooseLM StudioMsty StudioCherry StudioCursorVS Code + GitHub CopilotVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantClaude CodeCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatRaycast

All apps, with pros and cons →

Related guides

What it can do

Runs on your computer; the tool list appears in your AI app after you install it.

Basic checksi

We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 19, 2026

  • ✓ Pass

    Maker matches the repo

    The registry name was verified with the GitHub account suneel190700, which owns the repo

  • ✓ Pass

    Still maintained

    Code last changed 3 months ago

  • ✓ Pass

    Has a licence

    Uses MIT

  • – Not applicable

    Endpoint responds

    This one runs on your computer; there is no internet endpoint

  • – Not applicable

    Sign-in with OAuth

    Runs on your computer and uses local settings instead

  • … Not checked yet

    No hidden instructions in tool descriptions

    Runs on your computer; we can only read tool lists from internet endpoints

Published vulnerabilities (OSV.dev)

No entries on OSV.dev for mendrift-mcp@0.1.2

Checks only this package version, not every dependency · checked Nov 16, 2026

What we check → · JSON

Makers: ask for a re-check or report wrong info
For makers: add a badge to your README

If you made this, paste this code into your README. The badge links back to this page.

[![Featured on PopNx MCP](https://mcp.popnx.com/badge/suneel190700-mendrift-mcp.svg)](https://mcp.popnx.com/en/servers/suneel190700-mendrift-mcp)

Discussion

Ask, share a tip or report that it does not work. Sorted by what people marked helpful.

No posts yet. Be the first to ask or share.

Posts here are written by users, not the PopNx team.

Sign in to post

User reviews

No reviews yet

Sign in to write a review

Compared with others in Otheri

  • More stars than 0% of the other 20653 servers in this category
  • Published to the registry more recently than 13% of the other 20653 servers in this category
  • Basic checks equal to or better than 99% of the other 20653 servers in this category

Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.

Common questions

Is Mendrift safe?

It passed 3 of 3 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.

What keys or codes does it need?

The registry does not list a required key, though some servers ask you to sign in on first use.

Does it run on my computer or in the cloud?

It runs on your computer. Your AI app starts it when needed, and data does not pass through the maker’s servers (unless it calls an online service itself).

Is it still maintained?

The GitHub repo last changed 3 months ago, and the latest version (0.1.2) reached the registry 4 months ago.

Which apps can use it?

Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. You also need Python (uv) on your computer.

What can I use instead?

In the same category: World Monitor, openstatus, Oh My Posh Validator, Strata, Jitsu

Similar alternatives

Same category, other makers, commands already checked. Official ones first, then by stars.

Reading a card:↓ Downloads per weeki★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi

World Monitor

Real-time global intelligence on conflict, markets, energy, trade, shipping and cyber. 100 tools.

by koala73Widely used★ 90.9KUpdated 3 days ago

Runs in the cloud✓ service owner · worldmonitor.app

openstatus

Manage monitors, status pages, incidents and maintenances in your openstatus workspace.

by openstatusHQWidely used★ 9.3KUpdated 2 months ago

Runs in the cloud✓ service owner · openstatus.dev

Oh My Posh Validator

Validate oh-my-posh configurations and segment snippets against the official schema.

by JanDeDobbeleerWidely used★ 23.8K Stars for the whole repoUpdated 30 days ago

Runs in the cloud✓ service owner · ohmyposh.dev

Strata

MCP server for progressive tool usage at any scale (see https://klavis.ai)

by Klavis-AIGrowing★ 5.8KUpdated 14 months ago

Runs in the cloudRepo archived✓ klavis.ai

Jitsu

Manage Jitsu data pipelines: destinations, streams, connections, functions, live events.

by jitsucomWidely used★ 5.1KUpdated 4 months ago

Runs in the cloud✓ service owner · jitsu.com
Compared with similar ones
This one
Mendrift
World MonitoropenstatusOh My Posh ValidatorStrataJitsu
Runs whereInstalls on your computerRuns in the cloud, nothing to installRuns in the cloud, nothing to installRuns in the cloud, nothing to installRuns in the cloud, nothing to installRuns in the cloud, nothing to install
Needs a keyNoNoNoNoNoNo
Add with + in ClaudeNoNoNoNoNoNo
Basic checks3/34/44/44/41/44/4
LicenseMITAGPL-3.0AGPL-3.0MITApache-2.0MIT
PriceNo cost listed in the registryNo cost listed in the registryNo cost listed in the registryNo cost listed in the registryNo cost listed in the registryNo cost listed in the registry

README from the maker

Open the README

Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.

Mendrift

Autonomous MLOps incident response agent, plus mendrift-mcp — an open-source MCP server for drift detection and ML incident tooling.

pip install mendrift-mcp     # or: uvx mendrift-mcp

Published on PyPI and the MCP Registry as io.github.suneel190700/mendrift-mcp.

▶ [Live demo](https://mendrift-demo.onrender.com) — run a real incident in your browser: supply an alert, watch the agent diagnose it against a real MLflow registry, and approve or reject the rollback at the human-in-the-loop gate. Toggle between a crafted synthetic scenario and real US consumer-credit benchmark data. React frontend on a FastAPI backend; the free tier sleeps, so the first load may take ~40s.

When a production model drifts or degrades, Mendrift detects it, diagnoses the root cause from monitoring and registry evidence, proposes a remediation, and executes it only after human approval.

alert ──> classify ──> diagnose (MCP tools) ──> propose

│ │ noise ──> close human approval gate │ execute ──> verify recovery

Built with LangGraph (agent orchestration), LangChain (ChatAnthropic + bind_tools), the Model Context Protocol, Evidently, MLflow, and Claude (Haiku + Sonnet).

mendrift-mcp tools

tooltypepurpose
get_drift_reportreadper-feature drift distances + schema changes (Evidently)
summarize_metric_anomaliesreadproduction vs previous model scored on current traffic
get_deployment_historyreadregistry version transitions and aliases
diff_deploymentsreadparams / metrics / feature-schema diff between versions
propose_rollbackreadgenerates a reviewable rollback plan
execute_rollbackgatedrequires a single-use HMAC approval_token
open_incidentwriteincident record with diagnosis + evidence

Safety model

The approval gate is enforced in the tool layer, not the prompt: execute_rollback verifies a single-use, action-scoped HMAC token minted only by the human review flow — the minting function is never exposed over MCP. A prompt-injected or confused agent cannot execute writes.

Tested live: Claude was first ordered to roll back "with full authorization" (it proposed but declined to fabricate a token), then handed a fabricated token, which the gate rejected by constant-time HMAC comparison:

See tests/test_approval_gate.py, including the action-scoping test: a token minted for one model/version is invalid for any other.

Human-in-the-loop, crash-proof

The incident graph halts before execution (interrupt_before) and checkpoints every step to SQLite. The process can die; a new process resumes the same incident by thread_id after a human mints the approval token — which enters state only via update_state(), from outside the graph. Denial is a first-class path: no token → closed_approval_denied, no execution.

Agent design

stepmodelwhy
classifyHaikusingle constrained label; cheapest path
diagnoseSonnetmulti-hop tool reasoning over evidence
verifyHaikuthreshold check on fresh metrics

Routing lives in a code table (ROUTER_TABLE), not prompts, so cost per path is measurable config — ~3.9K input / 630 output tokens per incident. The diagnose loop is bounded (max 8 tool calls) with per-call retries and capped backoff; on tool failure the model receives a structured error record, and on budget exhaustion the agent degrades to an incident with partial evidence — it never invents a diagnosis. Destructive actions require affirmative evidence: a rollback is recommended only when retrieved evidence links the symptom to a specific deployment, never on deploy-correlation alone. The agent can also recommend monitor — real but mild, non-actionable drift is watched, not acted on.

Live mode

MENDRIFT_DEMO=0 runs the agent against real infrastructure rather than fixtures:

  • scripts/seed_demo.py trains two sklearn versions into a local MLflow registry — v13 clean, v14 with a schema swap and a training window polluted by missed-fraud labels (recall 0.72 → 0.18, AUC 0.84 → 0.82) — and writes reference/current frames
  • get_drift_report runs Evidently's DataDriftPreset over those frames, returning real Wasserstein/JS distances against per-metric thresholds, plus schema changes derived from actual column sets
  • get_deployment_history / diff_deployments read the registry and the underlying runs — real aliases, params, metrics
  • summarize_metric_anomalies scores the current window with both the production and previous versions, so it reports model divergence rather than population drift — a rollback clears it, ordinary data shift does not
  • an approved execute_rollback moves the production alias for real
uv run mlflow server --host 127.0.0.1 --port 5001        # separate terminal

PYTHONPATH=src uv run python scripts/seed_demo.py

rm -f demo.db MENDRIFT_DEMO=0 PYTHONPATH=src uv run python scripts/demo_interrupt.py start MENDRIFT_DEMO=0 PYTHONPATH=src uv run python scripts/demo_interrupt.py approve

A live run diagnoses from computed evidence — e.g. *"v2 introduced a schema swap replacing promo_flag with promo_flag_v2 … label_noise 0.0 → 0.45 collapsing val_recall 0.724 → 0.176 … 79.7% prediction-rate divergence from the prior version, model-induced, not population drift"* — then halts for approval and resolves.

The eval suite deliberately stays on fixtures: evals need determinism and zero cost in CI, while live mode exercises the real stack.

Live web demo (two worlds)

A hosted web app wraps live mode behind a browser UI: a React (Vite) frontend on a FastAPI backend, deployed on Render. A visitor submits an alert, the frontend posts it to /api/diagnose, and the backend runs the real LangGraph agent — live Claude reasoning over an embedded MLflow registry (sqlite://, seeded on boot) — then halts at the HMAC gate. Approve or reject and the backend resumes the graph via /api/decision, executing a real alias rollback and verifying recovery. The Anthropic key lives only on the server; runs are rate-limited since each calls a real model. Try it: [mendrift-demo.onrender.com](https://mendrift-demo.onrender.com).

The dashboard toggles between two seeded worlds, so the same agent can be seen against both a crafted scenario and genuine real-world data:

  • Synthetic (scripts/seed_demo.py, model fraud-scorer) — the crafted schema-swap incident: clean, teachable, an unambiguous rollback story.
  • Real US credit (scripts/seed_real.py, model credit-risk) — the Give Me Some Credit dataset (real US consumer-credit records, target SeriousDlqin2yrs) split by borrower age into reference/current windows for genuine feature drift, with a controlled model regression injected into v2 (asymmetric missed-default label noise) so the incident has ground truth. Real distributions and real Evidently drift; a known correct action. Measured gap: val_recall 0.637 → 0.156, AUC 0.854 → 0.810.

Injecting a known regression into real data is standard practice for validating a drift-detection system — it gives the evaluator ground truth for what the agent *should* decide while the drift computation still runs on genuine distributions.

The backend routes each request to the right world (model + parquet frames + label column) per the dataset field; the tool layer reads those from env vars, applied per-request under a lock so concurrent requests stay isolated.

Run the web app locally:

# 1. build the React frontend (FastAPI serves the built assets)

cd frontend && npm install && npm run build && cd ..

2. seed both worlds, then start the backend (frontend + API on one port)

export ANTHROPIC_API_KEY=sk-ant-... export MLFLOW_TRACKING_URI="sqlite:///$(pwd)/mlflow.db" PYTHONPATH=src uv run python scripts/seed_demo.py # synthetic world (fraud-scorer) PYTHONPATH=src uv run python scripts/seed_real.py # real world (credit-risk) PYTHONPATH=src uv run uvicorn app.main:app --port 8000 # open http://localhost:8000

For frontend development with hot reload, run cd frontend && npm run dev (port 5173); Vite proxies /api to the backend on port 8000.

Evaluation

src/mendrift/evals/ replays synthetic incident trajectories against the real graph — only the LLM (scripted) and the read tools (fixture world) are faked; the gated action tools are the genuine implementations, so the HMAC gate is exercised by every test. Four assertions per trajectory:

checkmeaning
no_ungated_writesevery execute_rollback carried a valid HMAC token — hard fail
classification_oktriage label matched
tool_sequence_okrequired tool calls occurred in order (extras allowed)
action_okterminal outcome matched

19 logic-distinct incident scenarios spanning the decision space, each with its own evidence shape and correct action:

  • Rollback — deploy-correlated drift or quality regression with affirmative diff evidence
  • Retrain — label/concept shift, segment-specific degradation (no valid rollback target)
  • Monitor — mild seasonal drift, low-importance-feature drift, holiday effects
  • Incident (investigate) — upstream schema rename, feature-store change, docs-only deploy, calibration break, threshold shift, silent data-quality drop
  • Graceful degradation — evidence tools down → incident with partial evidence, never a fabricated diagnosis
  • Noise — flapping / auto-resolved alerts closed with zero tool calls
  • Human-declined — well-founded rollback the reviewer rejects → closed, no execution

Scripted for fast CI, live for the measured rate:

PYTHONPATH=src uv run python scripts/run_traj.py --all          # scripted, fast

PYTHONPATH=src uv run python scripts/run_traj.py --all --live # real models

Live-model eval runs at ~95% task-success; the handful of run-to-run divergences reflect LLM eval variance on decision-margin scenarios. The live suite surfaced real failure classes during development — a JSON extractor masking a correct decision, a classifier baited by an alert's reassuring wording, and a diagnoser proposing rollback on correlation alone — each fixed at its own layer (parser, alert wording, evidence-rule prompt).

Quickstart (demo mode)

uv sync

MENDRIFT_DEMO=1 uv run mendrift-mcp # stdio MCP server with fixture data PYTHONPATH=src uv run pytest -v # gate + trajectory suite

Claude Desktop config:

{"mcpServers": {"mendrift": {

"command": "uvx", "args": ["mendrift-mcp"], "env": {"MENDRIFT_DEMO": "1"} }}}

Status

  • mendrift-mcp server over stdio, verified in MCP Inspector and Claude Desktop
  • seven tools with a read / gated / write permission taxonomy
  • HMAC-gated rollback with action-scoped single-use tokens (tests first)
  • LangGraph incident graph: SQLite checkpointing + human-approval interrupt, kill-resume proven
  • LLM nodes on LangChain (ChatAnthropic.bind_tools): Haiku classify/verify, Sonnet diagnose loop
  • 19-scenario trajectory eval across the decision space; ~95% live, zero ungated writes
  • CI: gate + trajectory suite o

Only the beginning is shown; read the rest in the repo