# Probe Research

> Probe Research is the system of record for a team's ML work — projects, experiments, runs, metrics, artifacts, notes and coding-agent sessions, with semantic search over all of it.

- [Probe Research](https://docs.prbe.ai/index.md): The system of record for everything your team does in ML — runs, metrics, files, decisions and the coding sessions around them.
- [Installation](https://docs.prbe.ai/installation.md): Install, update and remove Probe — through the wizard, with the commands underneath.
- [Quickstart](https://docs.prbe.ai/quickstart.md): Set Probe up with the wizard, then record a run you can open in the dashboard.
- [Core concepts](https://docs.prbe.ai/concepts.md): Workspaces, projects, experiments, runs, artifacts, notes — what each one is for and how they nest.
- [The wizard](https://docs.prbe.ai/wizard/overview.md): One command that installs, imports, configures, diagnoses and removes Probe on this device.
- [Install Probe](https://docs.prbe.ai/wizard/install.md): The guided install, screen by screen: pick your agents, review what gets set up, install, import.
- [Import research work](https://docs.prbe.ai/wizard/import.md): Bring in saved coding sessions and folders of existing project work, with a review step before anything uploads.
- [Existing imports](https://docs.prbe.ai/wizard/imports.md): Monitor background imports, read their results, resume what was interrupted.
- [Settings](https://docs.prbe.ai/wizard/settings.md): Automatic updates, research plan review and pasted images, and what a new session does by default.
- [Accounts](https://docs.prbe.ai/wizard/account.md): Sign in, switch between accounts saved on this device, and sign out.
- [Diagnose and repair](https://docs.prbe.ai/wizard/diagnose.md): probe doctor, the manual command list, and removing Probe from a device.
- [Unattended use](https://docs.prbe.ai/wizard/flags.md): Every wizard capability as a flag, for CI, fleet provisioning and headless machines.
- [Projects and experiments](https://docs.prbe.ai/tracking/projects.md): Create, organise, tag, nest, freeze and delete the structures runs live in.
- [Runs](https://docs.prbe.ai/tracking/runs.md): Open, hold, fork, wrap and close a run — and what happens when the process dies instead.
- [Metrics](https://docs.prbe.ai/tracking/metrics.md): Curves, scalars and breakdowns — and how to pick the shape before you log the first point.
- [Artifacts and files](https://docs.prbe.ai/tracking/artifacts.md): Record files against a run, experiment or project — uploaded or by reference — and read the team's Shared folder.
- [Datasets, sources and checkpoints](https://docs.prbe.ai/tracking/datasets.md): Record the data your models trained on, where it came from and every step it went through, and find every checkpoint your team saved.
- [Code capture and reproduction](https://docs.prbe.ai/tracking/snapshots.md): Record what was on disk when a run started, and rebuild it later with verification.
- [Spans, trajectories and trials](https://docs.prbe.ai/tracking/trajectories.md): Record what happened inside a run — rollouts, turns, tool calls — and capture Harbor trial directories.
- [Notes and documents](https://docs.prbe.ai/tracking/notes.md): The three homes for prose, the checkout/push editing loop, sub-notes, and the shared team note.
- [Summaries and Overview pages](https://docs.prbe.ai/tracking/summaries.md): The page each project and experiment keeps for teammates: who writes it, how to edit it, how your own blocks stay put, and its history.
- [Papers](https://docs.prbe.ai/tracking/papers.md): Record the literature a research project was built from — what each paper says, the repo read alongside it, and where the two disagree.
- [SDK quickstart](https://docs.prbe.ai/sdk/quickstart.md): Three lines in a training script, and what they do that a global does not.
- [API reference](https://docs.prbe.ai/sdk/reference.md): Client and Run methods, what each one writes, and the endpoint behind it.
- [Framework integrations](https://docs.prbe.ai/sdk/integrations.md): Miles, passive push, and wiring Probe into a trainer that already has a tracker.
- [CLI overview](https://docs.prbe.ai/cli/overview.md): Grammar, how to address an entity, and why most writes return before the server answers.
- [Authentication and accounts](https://docs.prbe.ai/cli/auth.md): Signing in, switching accounts and signing out — through the wizard, and the scriptable equivalents.
- [Command reference](https://docs.prbe.ai/cli/commands.md): Every probe command group and subcommand, with what it does.
- [The outbox](https://docs.prbe.ai/cli/outbox.md): Writes queue locally and survive a crash. How to watch delivery, and what to do when it fails.
- [Coding agents](https://docs.prbe.ai/agents/overview.md): Two plugins, one switch — what your coding agent gains when Probe is installed.
- [The Probe daemon](https://docs.prbe.ai/agents/daemon.md): A companion agent that runs beside your coding agent, records its work in Probe and brings it the team's prior work.
- [Skills](https://docs.prbe.ai/agents/skills.md): Every skill the plugin ships, when it fires, and what it makes the agent do.
- [Keep folders out of Probe](https://docs.prbe.ai/agents/probe-config.md): A gitignore-style .probe.config sets what Probe may read and record for each path, from any session, wherever it started.
- [MCP server](https://docs.prbe.ai/agents/mcp.md): Six read-only tools, token budgets, pagination, and the entity view matrix.
- [Session capture](https://docs.prbe.ai/agents/capture.md): What leaves your machine, who can see it, and every way to stop it.
- [Status line](https://docs.prbe.ai/agents/statusline.md): A one-line indicator in Claude Code showing whether this session is tracked.
- [Capy](https://docs.prbe.ai/agents/capy.md): Install Probe for Capy like any coding agent: capture the threads you start, and add Probe inside Capy.
- [The dashboard](https://docs.prbe.ai/dashboard/overview.md): Where the recorded work becomes browsable — a tour of every surface.
- [Projects](https://docs.prbe.ai/dashboard/projects.md): The directory of work: kinds, the project page, the Metrics center, papers, subprojects and published pages.
- [Experiments and runs](https://docs.prbe.ai/dashboard/experiments-runs.md): Experiment comparisons and conclusions, run detail, lineage, spans, trials, logs and sessions.
- [Charts](https://docs.prbe.ai/dashboard/charts.md): Curves, stat tiles, expression panels, zoom, and the captured rows behind a chart.
- [Data](https://docs.prbe.ai/dashboard/data.md): The team's data catalog: datasets, checkpoints and sources, where each came from, and the files on every project, experiment and run.
- [Search and knowledge](https://docs.prbe.ai/dashboard/search.md): Search by name, Ask by meaning, filter operators, and what the index actually covers.
- [Assistant](https://docs.prbe.ai/dashboard/assistant.md): Ask questions of your lab's history, and review the changes it proposes.
- [Workspaces, team and settings](https://docs.prbe.ai/dashboard/workspaces.md): Shelves for projects, each computer's workspace, who may edit what, members, devices, AI agents and settings.
- [Integrations](https://docs.prbe.ai/integrations/overview.md): GitHub, Weights & Biases and Benchling — live sync, backfill, and where imported content lands.
- [Weights & Biases](https://docs.prbe.ai/integrations/wandb.md): Import local run directories or a hosted project, and what source-backed runs keep at W&B.
- [GitHub](https://docs.prbe.ai/integrations/github.md): Send selected repositories to team Knowledge, and attach repos to projects for a read-through commit timeline.
- [Benchling](https://docs.prbe.ai/integrations/benchling.md): Bring notebook entries from selected Benchling projects into a named team workspace.
- [Backfill](https://docs.prbe.ai/integrations/backfill.md): Import work you have already done — a folder, saved coding sessions, or both.
- [Architecture](https://docs.prbe.ai/deploy/architecture.md): How Probe is put together, and which constraints decided the shape.
- [Self-hosting](https://docs.prbe.ai/deploy/self-host.md): Install Probe on your own Kubernetes cluster — the data-residency topology, including air-gapped.
- [Environment variables](https://docs.prbe.ai/reference/environment.md): Every variable the SDK, CLI and MCP read, and what each one overrides.
- [Troubleshooting](https://docs.prbe.ai/reference/troubleshooting.md): The failures that look like something else, and what to check first.
- [Glossary](https://docs.prbe.ai/reference/glossary.md): Every term Probe uses, in one sentence each.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.