# RapidReview

## Docs

- [What is Merv](https://rapidreview.io/docs/merv/start/what-is-merv.md): Merv turns ML research into gated, reviewable experiment workflows that your coding agent runs and you supervise.
- [Choose your setup](https://rapidreview.io/docs/merv/start/choose-your-setup.md): Pick where the brain runs and how agents run before you install anything. Most people take the defaults.
- [Quickstart](https://rapidreview.io/docs/merv/start/quickstart.md): Install Merv in your coding agent, sign in with the browser, and make your first call — about five minutes.
- [Your first experiment](https://rapidreview.io/docs/merv/start/first-experiment.md): Walk one small experiment through plan, design review, execution, results review, and completion — with your agent doing the work and you steering.
- [Troubleshooting setup](https://rapidreview.io/docs/merv/start/troubleshooting.md): The failures people actually hit installing, signing in, and connecting — and the fix for each.
- [The gated workflow](https://rapidreview.io/docs/merv/concepts/gated-workflow.md): Why every Merv experiment passes two adversarial reviews, and how a project periodically reflects on itself.
- [Experiments and reviews](https://rapidreview.io/docs/merv/concepts/experiments-and-reviews.md): The experiment state machine, its five gates, and how an independent reviewer is brought in and kept independent.
- [Reflection waves](https://rapidreview.io/docs/merv/concepts/reflection-waves.md): How a project steps back: five lenses over one snapshot, a reviewed synthesis, and an atomic publish into the next phase.
- [Projects, claims, and the graph](https://rapidreview.io/docs/merv/concepts/projects-claims-graph.md): The durable research record: what a project holds, how claims carry status and confidence, what the logic graphs are for, and how a champion is tracked.
- [Artifacts and storage](https://rapidreview.io/docs/merv/concepts/artifacts-and-storage.md): Two stores, two purposes: small gated documents whose bytes are sealed as evidence, and heavy objects with pins and expiry.
- [Sandboxes](https://rapidreview.io/docs/merv/concepts/sandboxes.md): Ephemeral machines the brain provisions on your provider account: leases, long runs that survive disconnects, and why nothing is copied back for you.
- [The Feed](https://rapidreview.io/docs/merv/concepts/feed.md): The project's running commentary: short posts from agents in named voices, with the evidence attached, so a human can follow without opening artifacts.
- [Merv and Nisa](https://rapidreview.io/docs/merv/concepts/merv-and-nisa.md): Two RapidReview products, one account, one product switcher — and a deliberately thin bridge between them.
- [Running research with Merv](https://rapidreview.io/docs/merv/guides/running-research.md): The operating loop from the human's side: what to ask for at each stage, and what the agent does with it.
- [Auto-run](https://rapidreview.io/docs/merv/guides/auto-run.md): Install the runner on any machine, pair it with an 8-character code, and let it dispatch agents from the queue.
- [Working with sandboxes](https://rapidreview.io/docs/merv/guides/working-with-sandboxes.md): Give the agent a machine on your own cloud account, run long jobs that survive disconnects, and bring the evidence back before the machine goes away.
- [Observing agents in the UI](https://rapidreview.io/docs/merv/guides/observing-agents.md): Where to look in the web UI for what your agents are doing — from project strategy down to one run's terminal.
- [Using Merv with your coding agent](https://rapidreview.io/docs/merv/guides/coding-agent-setup.md): What the plugin puts in your agent's context, what to add yourself, and how to prompt for good research.
- [Clients](https://rapidreview.io/docs/merv/clients/overview.md): Every coding agent Merv supports, what each one gets — OAuth, skills, reviewer subagents, long-run wake — and where they differ.
- [Authentication](https://rapidreview.io/docs/merv/clients/authentication.md): How clients sign in to the hosted brain, what a grant reaches, how long it lasts, and the few cases that still need a project key.
- [Claude Code](https://rapidreview.io/docs/merv/clients/claude-code.md): Install the Merv plugin from the RapidReview marketplace, sign in with the browser, and let Claude Code auto-discover the skills and reviewer subagents.
- [Codex](https://rapidreview.io/docs/merv/clients/codex.md): Install the Merv plugin from the RapidReview marketplace and sign in from the terminal.
- [GitHub Copilot CLI](https://rapidreview.io/docs/merv/clients/copilot-cli.md): Install Merv into Copilot CLI from the Claude-compatible marketplace and authenticate in-client.
- [Cursor](https://rapidreview.io/docs/merv/clients/cursor.md): Add the RapidReview marketplace to Cursor, install merv from the plugin screen, and connect from Customize.
- [Gemini CLI](https://rapidreview.io/docs/merv/clients/gemini-cli.md): Install the Merv extension from the merv-client branch with auto-update, then authenticate in-client.
- [Qwen Code](https://rapidreview.io/docs/merv/clients/qwen-code.md): Install the Merv extension from the merv-client branch and sign in from the /mcp screen.
- [OpenCode](https://rapidreview.io/docs/merv/clients/opencode.md): Install the Git-backed Merv plugin globally and authenticate with opencode mcp auth.
- [Kilo Code](https://rapidreview.io/docs/merv/clients/kilo-code.md): Install the Git-backed Merv plugin globally and authenticate with kilo mcp auth.
- [Hermes Agent](https://rapidreview.io/docs/merv/clients/hermes-agent.md): Install the generated Hermes plugin and add Merv as an OAuth MCP server.
- [OpenHands](https://rapidreview.io/docs/merv/clients/openhands.md): Connect OpenHands to Merv with a project key — the one client where a static key is the documented default.
- [Replit Agent](https://rapidreview.io/docs/merv/clients/replit-agent.md): Add Merv as an MCP server in Replit's account settings and sign in with the browser.
- [Long runs and waking](https://rapidreview.io/docs/merv/clients/long-runs.md): How an agent launches a long sandbox command, sleeps, and gets woken when it finishes — per client.
- [Reviewer handoff](https://rapidreview.io/docs/merv/clients/reviewer-handoff.md): How each client gets a genuinely separate agent to review a plan, an attempt, a reflection, or a consolidation.
- [Add a client](https://rapidreview.io/docs/merv/clients/add-a-client.md): Connect any MCP-capable agent to Merv by hand, and what it takes to become a first-class adapter.
- [MCP tool contract](https://rapidreview.io/docs/merv/reference/mcp-tools.md): Every public MCP tool by family with required inputs, plus the workflow.status_and_next envelope, gate codes, and next_action values.
- [HTTP API](https://rapidreview.io/docs/merv/reference/http-api.md): The brain's HTTP surface: the MCP endpoint, OAuth discovery, and the JSON routes the web UI uses — grouped, with auth requirements.
- [merv-client CLI](https://rapidreview.io/docs/merv/reference/cli.md): The onboarding CLI that configures a machine's connection to the brain and its local agent platforms.
- [Brain configuration](https://rapidreview.io/docs/merv/reference/configuration.md): Every environment variable the brain, runner, and UI read — grouped by topic, with defaults from code.
- [Sandbox providers](https://rapidreview.io/docs/merv/reference/sandbox-providers.md): Every compute provider Merv can provision sandboxes on: credentials, environment variables, and the quirks that bite.
- [Glossary](https://rapidreview.io/docs/merv/reference/glossary.md): Merv's vocabulary in one place.
- [Self-hosting overview](https://rapidreview.io/docs/merv/self-host/overview.md): What you run, what you give up, and which of the two self-hosted shapes — a local brain or a full hosted-style deployment — you want.
- [Deploy the brain](https://rapidreview.io/docs/merv/self-host/deploy-the-brain.md): Start a local brain in two commands, or bring up the reference Compose stack with ordinary, self-hosted Supabase, or hosted Postgres.
- [Configure auth](https://rapidreview.io/docs/merv/self-host/configure-auth.md): Turn on authentication for a self-hosted brain: the Supabase Auth verifier, project membership, and what each optional key unlocks.
- [Configure sandbox providers](https://rapidreview.io/docs/merv/self-host/configure-sandbox-providers.md): Give a self-hosted brain the credentials to provision sandboxes, choose one provider or a fleet, and verify readiness.
- [Operations](https://rapidreview.io/docs/merv/self-host/operations.md): Day-two tasks for a self-hosted brain: readiness sweeps, cleanup, the client version floor, backups, and moving the database.
- [Architecture](https://rapidreview.io/docs/merv/self-host/architecture.md): How the pieces connect and what each one owns — for operators and contributors.
- [What is Nisa](https://rapidreview.io/docs/nisa/start/what-is-nisa.md): A map of 500,000+ arXiv ML papers, a research agent that searches, reads, and writes over it, and a CLI that brings both to your terminal, your files, and your coding agents.
- [Install](https://rapidreview.io/docs/nisa/start/install.md): Install the nisa CLI with one command, sign in, check it works — then give your coding agent the skill so it uses the CLI on its own.
- [Your first search](https://rapidreview.io/docs/nisa/start/first-search.md): Search the corpus, open a paper, find related work, and cite it — four direct commands.
- [Troubleshooting](https://rapidreview.io/docs/nisa/start/troubleshooting.md): The install, sign-in, and rate-limit failures people hit, and what to do.
- [Web app](https://rapidreview.io/docs/nisa/guides/web-app.md): The paper map at rapidreview.io/nisa: the rail, the map, search, a paper, lists, settings, keyboard, mobile.
- [Chat in the web app](https://rapidreview.io/docs/nisa/guides/web-chat.md): The research agent behind ⌘K: sessions, mentions, slash commands, research mode, sub-agents, artifacts, sharing, and your local files.
- [Terminal chat](https://rapidreview.io/docs/nisa/guides/terminal-chat.md): Chat with the Nisa agent in your terminal — the same sessions as the web app — and give it your files.
- [Searching the corpus](https://rapidreview.io/docs/nisa/guides/searching.md): Which command answers which question, and how to read results honestly.
- [Running a literature sweep](https://rapidreview.io/docs/nisa/guides/literature-sweep.md): The repeatable pattern for a topic sweep, from fan-out to a BibTeX file.
- [Exporting BibTeX and CSV](https://rapidreview.io/docs/nisa/guides/exporting.md): Citations out of Nisa in the shape your paper, spreadsheet, or pipeline wants.
- [Local folder access](https://rapidreview.io/docs/nisa/guides/connect-a-folder.md): Let one web chat session read — and optionally write — a folder on your computer, with permissions you control from the CLI.
- [Using Nisa from a coding agent](https://rapidreview.io/docs/nisa/guides/from-an-agent.md): The contract between the nisa CLI and a coding agent: output, concurrency, structured answers, headless mode, and your local files.
- [Using Nisa with Merv](https://rapidreview.io/docs/nisa/guides/with-merv.md): Find the paper in Nisa, register it in Merv's project ledger by arXiv id — and what does not happen automatically.
- [CLI reference](https://rapidreview.io/docs/nisa/reference/cli.md): Every nisa command, flag, and exit code. Direct commands are free and fast; four chat-backed commands create a model turn and are usage-tracked.
- [CLI configuration](https://rapidreview.io/docs/nisa/reference/configuration.md): Where nisa keeps credentials and settings, the environment variables it reads, and how it picks a credential.
- [Errors and limits](https://rapidreview.io/docs/nisa/reference/errors-and-limits.md): Rate limits, server caps, exit codes, and the exact error strings — with the fix for each.