Questions, answered

Kuickr hosts a folder of your docs at clean URLs. Point it at a docs folder — from your repo, the CLI, or an AI chat — and every file gets a permanent address like kuickr.co/you/handbook/intro.md. Markdown renders to clean pages; HTML, PDFs and images are served as-is.

Most "host this" tools give you a link — a random URL to one throwaway page. Kuickr gives you a place. Pages live in folders that own a readable path (kuickr.co/you/handbook/intro.md), so you find them by where they live, not by remembering a slug. The same agent — or you, three weeks later — can read, edit, restyle, search, and see the full revision history. You publish and keep, instead of publish and forget.

1. Open Settings → Connectors → Add custom connector.
2. Name it kuickr and set the URL to https://kuickr.co/mcp. Leave OAuth Client ID and Secret blank — Kuickr registers the client automatically (Dynamic Client Registration), so you don't need them.

3. Click Add. Claude opens a browser to sign in to Kuickr — approve the connection (no token to paste).
4. The connector shows as connected with its 10 tools. Set each tool to allow or ask — e.g. keep kuickr_space / kuickr_folder on "ask," the rest on "allow."

Now in any chat: "Use kuickr to list my pages" or "Read my-org/handbook/intro.md and fix the typo in the first heading."

Some older Codex CLI builds require the remote MCP OAuth client to be enabled before codex mcp login kuickr works. Add experimental_use_rmcp_client = true to ~/.codex/config.toml, or run the login once with codex -c experimental_use_rmcp_client=true mcp login kuickr. Then open a new Codex chat and run /mcp to confirm Kuickr is connected.

Kuickr exposes an MCP server at https://kuickr.co/mcp with 10 tools to list, read, search, choose design tokens, create, edit, move, and share pages and folders. Creating a page at an existing path edits it — including HTML pages, hosted as-is. Every tool is permissioned, and you approve them per-tool the first time you connect.

Once connected, just talk to it: "Use kuickr to list my pages." · "Publish these notes to kuickr and give me the link." · "Read you/handbook/intro.md and fix the typo in the first heading." · "Search my kuickr pages for the deploy runbook."

Yes — per folder. In a folder's settings there are two toggles: AI can read and AI can update. Turn off "read" to keep a folder out of agent reach entirely; turn off "update" to make it read-only to agents (they can list and read it, but create/edit/share is blocked). Both default on, so nothing changes until you flip them.

No — and this matters. CLI push writes from your files; chat/agent edits write to Kuickr directly. They don't flow back to your repo, and the next push from your repo would overwrite them. Pick one source of truth per folder: repo-driven (push only) or Kuickr-driven (edit in place via chats).

curl -sSL https://kuickr.co/install.sh | sh

macOS & Linux. Or brew install nauman/tap/kuickr. Then run kuickr login --browser and approve the browser prompt. See the full CLI page.

That page is the browser-login approval step. When you run kuickr login --browser, the CLI starts a temporary callback on your own computer, usually something like http://127.0.0.1:59429/callback. Approving sends an API token back to that local callback so the CLI can save it. Only approve if you just ran the command yourself. Kuickr rejects non-loopback callback hosts, so the token cannot be redirected to a remote website. The step-by-step version is on the CLI login guide.

kuickr docs create handbook      # scaffold a starter folder
kuickr attach handbook --space you --dir handbook
kuickr open you/handbook

The folder structure becomes your URL structure — numbered files order the sidebar, and ##/### headings build the on-page table of contents automatically.

No. kuickr push is a content sync, not a git operation. It walks your --dir folder, hashes every file, and uploads. The server keeps what changed and skips what didn't — there's no git history involved. The --repo/--branch flags on attach only record where the docs came from as a label; they don't pull from git.

Every matching file (.md, .html, .pdf, images) is uploaded with a content hash. The server upserts changed files, skips unchanged ones, and prunes files you've deleted locally — so the hosted folder mirrors your directory. Pass --keep to skip pruning.

Edit the file and re-run kuickr push. To automate it, wire kuickr push into a git hook or CI step (e.g. a GitHub Action on merge) so every commit re-syncs.

Private by default. A page is only reachable anonymously once you share it. Pages an agent drops in your inbox are private until you share them.

From your dashboard, use Share on a folder — or have an agent call the share tool — to mint a public link, optionally password-gated. Share a folder once and every page under it is reachable; no re-sharing per page.

The easiest path is kuickr login --browser. If you need manual or scripted auth, copy a token from settings and run kuickr login --token <TOKEN>. The CLI stores it in ~/.kuickr/config.toml. For agent sessions, export KUICKR_TOKEN (and KUICKR_HOST) to override the file — handy with a secret manager. Codex users can point bearer_token_env_var = "KUICKR_TOKEN" at it in ~/.codex/config.toml.