ADoorYour data account Get your account

Track eval results over time

Every score, for every model and checkpoint, on every dataset and harness version. The eval job writes with add-only access, so old scores can’t be quietly changed.

Use cases All use cases

For builders

How to connect ChatGPT and Claude to the same data How to give an AI agent safe, scoped access to data A database for vibe-coded apps, with access built in

AI and engineering

Keep a daily work log Keep a prompt library for Claude, ChatGPT and Cursor Let agents build on each other’s findings Run a task queue for your agents Share agent traces across a team Share research notes Track eval results over time Start an experiment tracker Keep a model training log Keep feature flags and config agents can read Keep bug reproductions and test cases Record deploy and release history Keep an inventory of API keys and services Start a dataset registry Collect standup notes from assistants Keep an incident log Log benchmarks and latency measurements Track GPU and compute spend Start a labeling and annotation queue

Personal

Keep a side-project idea backlog Start a habit tracker Start a reading log Track what you’re learning Track subscriptions Start a budget tracker Start a movie and TV tracker Start a home inventory Keep a plant care log Keep a car maintenance log Plan your travel Keep a tasting journal Track a collection Keep notes on the people you know Track job applications Keep a vocabulary list Track important documents Keep a food log Start a training log Keep a log of your own numbers Keep a schedule of what is due and when to refill Start a recipe collection Keep a list of things to buy Share a pet care log Keep a daily journal Keep gift ideas and birthdays

Business

Track consulting time Track freelance clients and invoices Track fundraising Start a lightweight issue tracker Back an internal tool Store a customer-facing app’s data Plan a content calendar Collect customer feedback and feature requests Manage event attendees and sponsors Report from many sources Start a customer list Track sales deals Keep a customer support log Track hiring Start an employee directory and onboarding checklist Track inventory Track orders and suppliers Start an asset and equipment register Track partners and resellers Track contracts and renewal dates Keep a price list and build quotes

Collaborative

Split expenses Share a research reading list Plan an open-source project roadmap Build a shared world-building bible Plan a house renovation Start a shared household list Share a household chores rota Keep a family schedule Track kids’ activities Plan a trip with friends Plan a wedding or party Run a book club Run a sports team or club Run volunteer and community sign-ups Run a student group project Start a neighborhood tool-lending list Share a weekly meal plan Write babysitter and caregiver handover notes Plan a shift schedule Plan a carpool rota Share a packing list Coordinate help for a parent

An example

@ADoor Eval job finished for r-0412 at 40k: GSM8K 61.3, MATH 24.8, HumanEval 38.4. Eval harness v2.3.

ADoor Added 3 entries to Eval results

Eval results

ModelDatasetScoreHarness
r-0412@40kGSM8K61.3v2.3
r-0412@40kMATH24.8v2.3
r-0412@40kHumanEval38.4v2.3

What you can ask later

  • “Which checkpoint has the best GSM8K on harness v2.3?”
  • “Did anything regress more than two points since last week?”
  • “Show scores for every model on MATH, newest first.”
  • “Which results were run on the old harness and need rerunning?”
  • “Chart GSM8K and MATH scores for every checkpoint of r-0412.”

Who can see what

  • The eval job writes results with add-only access, so it can’t change old scores.
  • An agent like Cursor or Claude can read results and can’t write them.
  • Researchers read everything and add notes.
  • A partner lab can read results for one shared model only.

How it compares with a spreadsheet or an eval platform

Eval platforms like Braintrust or LangSmith are better at running the evals themselves, per-example traces and side-by-side output review.

As the long-term scoreboard, ADoor holds up: add-only results, plain SQL for regressions, and read-only access for the agents that decide what to train next.

How to start

  1. Open an assistant like ChatGPT, Claude, Grok or Cursor and connect ADoor through MCP.
  2. Tell it: “Set up eval results: Eval results (model, checkpoint, dataset, dataset version, score, harness version, run at).”
  3. Give the eval job add-only access, and share one model’s results with a partner lab if you need to.

Get your account

Questions people ask

Can results be changed after the fact?

Give writers add-only access, and the history records each change and who made it.

Can I store per-example outputs?

Yes, as entries in a linked list. For large volumes, store them elsewhere and link them.

Is ADoor free?

Yes. The free plan holds 3 tables and 100 rows. Pro is $10 a month per workspace, with 1 GB of storage and no per-GB charges.

What if someone leaves?

Remove their access and it stops working on their next request. The history still shows what they added and changed.

Updated October 20, 2026.

This page as Markdown