# Nproc Handbook

A concise reference for how Nproc is organized, how plans and billing work, how to operate the canvas, and how to integrate with Claude through MCP.

## Overview

- A workspace is a canvas (whiteboard).
- Issues and docs are scoped to a workspace.
- A user's access is defined entirely by the workspaces they belong to.
- A team contains multiple workspaces.

## Supported Browsers

- Recommended: latest Chrome.
- Safari also works for the core feature set.

## Account

### Sign-in

Sign-up and sign-in use Google authentication only. There is no email-and-password login.

### Email address

Your email address is tied to the Google account you signed up with and cannot be changed.

### Display name and avatar

Both are set per team, so you can present differently in each team.

### Account data export

Avatar menu > Account & privacy > Export my data. Exports are self-service, free, and delivered in a structured machine-readable format (such as JSON). One export per 24 hours. Each export stays downloadable for 7 days, then is deleted automatically.

### Account deletion

An account cannot be deleted while any of the following applies:

- A team where you are the only Admin with an Nproc account still exists. An invited Admin who has not created an account yet does not count.
- A Pro cancellation you registered has not taken effect yet (the team is still within the paid period).
- An account data export you requested is still in progress. This block clears automatically within 24 hours, once the export completes.

## Roles

Three roles exist: Admin, Member, Guest. Only Admins can invite users. Members and Guests cannot invite anyone.

### Admin

Sees the full list of workspaces in the team, but still needs to be invited (or invite themselves) to open a workspace. Only Admins can invite users to workspaces or manage billing.

### Member

Accesses only the workspaces they were invited to. Can belong to multiple workspaces within a team. Can see the list of those workspaces and the members invited to each.

### Guest

Same visibility rules as Member, but a Guest can belong to only one workspace per team.

## Team membership

### Role changes

Only Admins change roles, and one user at a time. Role changes are not available on the Free plan.

### Seat effect of a role change

Changing a Guest to Admin or Member uses one more seat, and is possible only while the number of seat-holding users stays within the purchased seat count. Changing an Admin or Member to Guest frees one seat.

### Removing a user from a team

The user loses access to the team and its workspaces. Issues and docs they created remain, including their name. The number of used seats adjusts automatically; the billed seat count does not - adjust it from the Billing page.

### Leaving a team

An Admin can leave unless they are the only Admin with an Nproc account in the team (an invited Admin who has not created an account yet does not count), or they registered a Pro cancellation that has not taken effect yet. Members and Guests can leave at any time.

## Invitations

- Admins invite users to a workspace. There is no separate team invitation; joining a workspace also makes the user a team member.
- A Guest can belong to only one workspace per team.

### Invitation email

An invitation email is sent only when the invited address has no Nproc account. It carries a sign-in link.

Within one team, an invited address receives one email. While that invitation still stands, the same address cannot be invited to that team again. If the person is removed from the team and invited again, a new email is sent. A different team inviting the same address sends its own email.

An address that already has an Nproc account receives no invitation email. That user is added immediately, and the team and workspace are there the next time they open Nproc.

Nproc caps how many invitation emails one Admin can trigger per day. Invitations past that cap still go through - only the email is skipped. Ask the invited person to open Nproc and sign in with the invited address. To have the email sent instead, remove the person from the team and invite them again once the cap lifts.

## Workspaces

### Active and Archived

Bookmarks are the only difference: only Active workspaces can be bookmarked. Everything else works the same. Admins and Members can archive a workspace and restore it to Active.

### Bookmarks

Per user. They are not shared with the team.

### Renaming

Any member of a workspace can rename it, regardless of role.

### Deleting

Admins can delete any workspace in the team; Members can delete workspaces they belong to. Deletion cannot be undone, and the workspace's issues and docs are deleted with it.

### Workspace export

Any member of a workspace can export it, regardless of role. Open the workspace, use the menu in
the top right of the canvas, and choose Export workspace. The file downloads directly in your
browser as a ZIP.

What the ZIP contains:

- canvas.json - the complete copy of the canvas, with item text kept as rich-text JSON so nothing
  is lost in conversion.
- workspace.md - the same canvas content in a readable form. Formatting is normalised.
- issues.json and issues.md - the issues in the workspace, with their comments and status changes.
- documents.json and documents.md - the documents in the workspace, with their body text.
- images/ - the image files placed on the canvas, and those used in the body text of issues,
  comments and documents.
- manifest.json and README.md - counts, metadata, and what is and is not included.

What it does not contain: past revisions, the text of comments you deleted, links you made by hand
(from a list, a menu, the right-click menu, or by dragging a pin onto the canvas), who has read
what, notification settings, and anything you deleted. Links written inside body text are part of
the text, so they remain. Archived items are included, because archiving is not deletion.

The ZIP can contain personal data of other people, including names of members and anyone mentioned
in the text. Handle it according to your organisation's policy.

## Plans

### Free

- Up to 1 Admin, 3 Guests, 3 workspaces.
- Archived workspaces also count toward the 3-workspace limit.
- At least one Admin is always required.
- Inviting Admins or Members is not available.
- Copy-paste across workspaces is not available. This covers the canvas, and images in issue, comment and document bodies: an image added while one workspace was selected cannot be saved into a body that belongs to a different one, whether you pasted it from elsewhere or switched workspace after adding it. Pasting within the same workspace is unaffected.

### Pro - $4 / month per paid user per team, plus applicable tax

- At least 1 Admin required.
- Unlimited Members, Guests, and workspaces.
- Guests remain limited to one workspace per team.
- Copy-paste across workspaces is available when every team involved is on an active Pro plan. This covers the canvas, and images in issue, comment and document bodies.
- Paid users = Admins + Members. The seat count you purchase caps the number of paid users you can invite. Guests do not consume seats.
- Billing is per team. The same user joining a second team counts as a separate paid seat there and is billed $4 / month again in that team, plus applicable tax. There is no cross-team discount or shared seat.

### Free Trial

- 14 days of the full Pro feature set.
- A credit card is required to start.
- There is no trial-specific seat limit.
- Pro capabilities are available, including unlimited workspaces.

## Billing

Invoices, credit-card updates, and retrying a failed charge are all done from the Billing page.

### Seat count changes

Seat count changes are made from the Billing page. Removing a member from the team does not automatically reduce the seat count (the billed quantity). The prorated adjustment is not charged immediately; it is added to or offset against the next billing cycle's invoice.

### Trial conversion

At the end of the trial, the team automatically converts to Pro. You can cancel at any time before conversion. If you cancel during the trial, the team does not convert. Canceling trials multiple times in a short period temporarily prevents starting a new trial.

### Failed payments

If a monthly charge fails, update the card on the Billing page and retry the payment from the same place.

If the initial payment for a team is not completed within the time window set by the payment provider, the team is locked.

### Locked teams

- A team becomes locked when its initial payment is not completed within 23 hours of starting the subscription.
- Lock scope is per team. Workspace access is suspended only in the locked team. Other teams you belong to are unaffected.
- The original subscription cannot be revived after the 23-hour window expires.
- An Admin can start a brand-new subscription from the locked team screen. The new subscription replaces the expired one.
- If the restart keeps failing, please contact support.
- A new team can be created at any time from the avatar menu, and a new trial can be started there. Members and content from the locked team are not migrated.
- Account data export (avatar menu > Account & privacy > Export my data) is account-scoped and remains available regardless of any team's lock state.
- If a charge appears on a card but the team is still locked, please contact support and include the receipt so the records can be reconciled.

### Payment provider cookies

The Billing page loads a script from our payment provider Stripe (Stripe.js), which sets fraud-prevention cookies (__stripe_mid / __stripe_sid). See section 12 of the privacy policy for details.

## Cancellation

- Nproc remains usable until the end of the current billing period.
- Once cancellation is registered, the seat count is locked.
- Outstanding balances are charged immediately at cancellation.
- Remaining credit rolls over to the next contract. If you do not intend to resubscribe, a refund request can be filed from the Billing page after cancellation.
- When the cancellation takes effect at the end of the billing period, all team members other than the cancelling Admin are removed from the team. The cancelling Admin retains access to the team's data.
- Canceling during the trial takes effect immediately. Pro features stop at that point, and all team members other than the cancelling Admin are removed from the team.

### Team deletion

Only an Admin can delete a team. A team cannot be deleted while the subscription is active, including during the free trial. Deleting a team permanently deletes the team's business data. Deleted data cannot be restored.

Export each workspace you want to keep before cancelling or deleting the team. Copies are not provided after deletion. Export runs per workspace and needs you to be a member of it, so join a workspace first if you are not.

## Issues

- Creating an issue requires a title. The description is optional. Any workspace member can create issues, regardless of role.
- The title can be edited by any workspace member. The description can be edited only by the issue creator. A comment can be edited only by its author.
- Descriptions and comments keep a change history that workspace members can view.
- An issue can be deleted by an Admin or the issue creator. A comment can be deleted by an Admin or the comment author.
- An issue is either open or closed. Any workspace member can change the status.
- Mentions are available in issue descriptions and comments.
- Type "/issue" in a description or comment to embed a reference to another issue in the same workspace. The reference shows "#number title" using the title at the time it was embedded, in full; it does not follow later title changes. Select it to open that issue. Removing the reference from the text removes it as a whole.
- Embedded references link the two issues. The link is created when the description or comment is saved, and removed when the reference is deleted and saved, or when the comment is deleted. The issue sidebar lists linked issues under "Issues". Links created from the text cannot be unlinked from that list - remove the reference in the text instead. Links added with the "+" in that list can be unlinked from the list. Past revisions may still show a reference, but they do not create links.
- Type "/doc" in a description or comment to embed a reference to a doc in the same workspace. The reference shows a doc icon followed by the doc title at the time it was embedded, in full. Select it to open that doc. Saving links the issue to that doc; the doc appears under "Docs" in the issue sidebar and cannot be unlinked from that list - remove the reference in the text instead.

## Docs

- Any member of the workspace a doc belongs to can create, edit, or delete that doc.
- Docs keep past revisions that workspace members can view.
- A doc can be linked to issues, and an issue to docs. The issue sidebar lists linked docs under "Docs" and the doc sidebar lists linked issues under "Issues". Links you add by hand from those lists can be unlinked from there.
- Type "/issue" in a doc to embed a reference to an issue in the same workspace. Saving the doc links it to that issue; removing the reference and saving unlinks it. Links created this way cannot be unlinked from the sidebar - remove the reference in the text instead.
- Type "/doc" in a doc to embed a reference to another doc in the same workspace. Saving the doc links the two docs. The doc sidebar lists the docs referenced from the text under "Docs" (the doc being referenced does not list the referencing doc), and lists notes and table rows whose text references this doc under "Canvas items". Neither list has an unlink action - remove the reference in the text instead. Selecting a "Canvas items" entry jumps to the canvas; if the item was deleted, the link is removed and you are told so.
- Deleting a doc also removes every link to and from it.

## Links in text

Applies to issue bodies, comments, docs, and notes.

- Pasting a URL when nothing is selected inserts a shortened label - the domain, then the last part of the address. The link still points to the full URL, which you can see by hovering over it.
- Pasting over text you selected keeps your text as the link label. Links already in the body are left as they are.
- Links are shown with a link icon and a tinted background.

## Search

- Search covers titles and bodies of docs, issues, and notes (including table rows), plus table titles and workspace names. Values inside table cells are not covered.
- All workspaces you belong to are searched, including archived ones. A search can also be limited to a single workspace.
- Keywords match partially, must be at least two characters, and are combined with AND when there are several.
- All keywords must appear together in one title or one body text. For an issue, the description and each comment count as separate body texts, so a keyword in the description and a keyword in a comment will not match together.
- Separate keywords with spaces. Symbols are treated as part of a keyword, so "Q&A" and "C++" are searchable, and "a,b" is one keyword rather than two.
- Docs, issues, workspace names, and the titles of notes and tables are searchable immediately. Note bodies take some time after an update before they are searchable.
- For links, only the text shown in the body is searchable. Link addresses are not covered.
- An embedded issue reference is searchable by its number ("#123"), not by the referenced issue's title.
- An embedded doc reference is searchable by the doc title at the time it was embedded (up to 24 characters); it does not follow later title changes.

## Notifications

- You are notified when a new issue's description mentions you, when a comment is added to an issue you are involved in, when an issue description or a comment is edited (people newly mentioned by the edit are notified as well; people already mentioned are not notified again), and when an issue is closed or reopened with a comment. A status change without a comment does not notify.
- Notifications go to the people involved in the issue - its creator, comment authors, and mentioned users - except the person who performed the action.
- Notifications appear as a badge in your browser. Mention notifications are marked so you can tell them apart from other updates.

## Collaborators

- The Collaborators page shows how active each person has been in the team. Each square is one day for one person, and the bar chart above them is the weekly total.
- Three actions count as one contribution each: creating an issue, adding a comment to an issue, and closing or reopening an issue with a comment attached.
- Nothing else counts. Changing an issue status without a comment, editing or deleting an issue or a comment, and any work in docs, notes, tables, or the canvas do not add to the count.
- A contribution is credited to the person who performed the action, including actions Claude performs on your behalf through the MCP integration.
- Counts only go up. Deleting an issue or a comment afterwards does not take the contribution back.
- Days are counted and displayed in UTC, not in your own time zone, so an action can land on a different square than your own calendar day suggests.
- A person's count covers everything they did in that team, across every workspace they belong to, not only the workspaces you share with them.
- Who appears depends on your role. An Admin sees everyone in the team. A Member or Guest sees only the people who share a workspace with them, and archived workspaces are excluded. The weekly bar chart totals only the people you can see.
- Someone who leaves the team, or is removed from it, disappears from the page, and their days drop out of the weekly totals as well. If they are invited back into the same team, their earlier history reappears with them. If the account itself is deleted, that history is deleted for good.

## Canvas

### Deleting an issue from the canvas

Deleting an issue shape removes it from the canvas only. The issue itself is not deleted - its description, comments and history stay intact, and it remains reachable from the Issues list. An issue shape is placed on the canvas at the moment the issue is created, and an existing issue cannot be placed on the canvas again, so the shape does not come back once it is deleted.

### Linking a canvas shape to an issue

Right-click a note, section, sticky, rectangle or table on the canvas and choose "Link to issue" to attach it to an issue in the same workspace. One shape can be linked to several issues. The issue shows its linked shapes under "Canvas items" in the sidebar, using the name the shape had when it was linked. Select an entry to jump to that shape on the canvas, or use its menu to unlink it. Deleting a shape does not remove the link - if you open a link whose shape has since been deleted, Nproc tells you so and removes that entry. Undo and redo restore or remove shapes only - they never re-create a link that was already removed.

There are two more ways to link, both of which work by position. Drag an issue pin onto a note, sticky, rectangle or table: the target lights up while the pin is over it, and releasing there creates the link. Or pick the issue tool and click on one of those shapes: the pin you create is linked to the shape under the click as soon as it is placed. If you cancel the form, no pin is created and nothing is linked. Sections cannot be linked this way - use the right-click menu for those.

A pin linked by either of those two ways follows its shape: move the shape and the pin moves with it. Moving the pin alone leaves the shape where it is. To unlink, drag the pin off the shape and release once the shape lights up in the warning colour; the pin stops following at the same time. Links made from the right-click menu or from the text of a note do not follow, because the pin's position carries no meaning in those cases. Undo restores the pin's position only - it does not undo the link or the following.

You can also link a note or a table row to a doc from inside its text: type "/doc" and pick a doc. The link is created right away and is listed under "Canvas items" in that doc's sidebar; deleting the reference removes it.

You can also link from inside the text of a note or a table row: type "/issue" and pick an issue. The reference appears as "#number title" and the link is created right away; deleting the reference removes the link. Links created from the text are listed under "Canvas items" too, but cannot be unlinked from that list - remove the reference in the text instead. For these text links, undo and paste follow the text: undoing a deletion restores the link along with the reference, and a reference pasted with a note is linked the next time that note is opened. Selecting a table row entry jumps to the table and opens that row.

### Copy and paste

Standard shortcuts - Cmd/Ctrl + C and Cmd/Ctrl + V. Notes and tables can be copied together in one operation. Issues cannot be copy-pasted - an issue shape is tied to the issue record itself, so it cannot be duplicated. Pasted shapes are not linked to any issue, even if the original was. Pasting within the same workspace is always available. Pasting into a different workspace requires an active Pro plan for every team involved, and you must be a member of each of them. This applies to another workspace in the same team as well.

When a paste is blocked, the team you need to upgrade is not always the team you are currently in, so the notice names the team to upgrade:

- Only the team you are pasting into is on Free: upgrade that team.
- Only the team you copied from is on Free: upgrade that team instead. Upgrading the team you are in does not lift the block.
- Both teams are on Free: both need an active Pro plan.

The notice does not link to a billing page. Open Billing for the team you want to upgrade from its own team menu.

### Image import

Supported directly onto the canvas. Up to 10 MB per image.

### PDF import

Imported as images, up to 100 pages per single upload operation. The canvas itself can hold more than 100 images - repeat the upload to add more. Each rendered page must also be within the 10 MB per-image cap.

## Appearance

Each user can tune display settings from the Appearance modal. Preferences are saved per user on the server, so the same settings follow you across devices and across teams. Language, Color mode and Contrast are also synced to browser cookies when you make an explicit choice (for Color mode and Contrast, choosing System removes the cookie), so they apply from the first paint. The modal shows a short notice about this. A Reset to default button appears at the top of the modal whenever any value differs from the default.

### Accessibility settings

#### Contrast

Normal, High, or System. High strengthens the contrast of text, buttons and borders in the shared interface. Normal keeps the default look. System is the default and follows your operating system setting for increased contrast, so if you have that turned on you get the stronger contrast from the first paint without changing anything here. Contrast works together with Color mode, so Light and Dark each have their own Normal and High.

- Note: if your operating system is running a forced high contrast theme (for example Windows contrast themes), the operating system replaces the colors itself and this setting makes no visible difference.
- Note: some individual screens are not covered yet, so the difference may be smaller in places.
- Note: heading colors and emphasis colors that are set to a color, including the starting heading colors described under Heading colors, stay the same in every mode. Only the built-in defaults change between Light, Dark, Normal and High.

#### Canvas contents button

Show or Hide. The Contents button in the canvas header opens a list of the canvas contents that works with the keyboard and screen readers. Issues, pen strokes, and lines and arrows without text on them appear as a count rather than as individual entries. Hide is the default. Hide only takes the button out of view - the button is still there and appears when you get to it with the Tab key. Choose Show to keep it visible at all times.

### App settings

#### Language

English or Japanese. Stored per user on the server.

#### Color mode

Light, Dark, or System (follows the OS).

#### Items per page

10, 30, or 50. Applies to workspace, issue, and doc lists.

### List settings

Applies to the workspace, issue and doc lists, and to the Contents, Issues and Docs panels opened from the canvas header.

#### Font size (lists)

Small, Medium, or Large. Sets the size of each row's title. The smaller supporting line under or beside the title (for example the issue number and who updated it, or the workspace a doc belongs to) follows at 2px smaller. Status badges and icons keep a fixed size, as do the dates in the panels opened from the canvas header. Dates in the workspace and issue lists follow the setting with the supporting line. Default Medium, which matches how lists looked before this setting existed.

- Note: these are a different set of sizes from the Editor Font size setting, which starts smaller. Lists are read by scanning titles, so the smallest list size is still comfortable to scan.

#### Spacing (lists)

Compact, Normal, or Relaxed. Sets how much room each row takes. Default Normal.

- Note: this replaces the earlier Workspace / Issue / Doc list density settings, which each controlled the same thing for one list only. One Spacing setting now covers all of the lists in this section, including Contents, which previously had no setting at all. If you had picked Compact for a list before, it now starts at Normal and you can choose Compact again here.

#### Issues list default tab

All, Open, or Closed. Sets which tab is already selected when you open an issue list. Applies to the issues page and to the Issues panel opened from the canvas header. Default All. Choose Open if you would rather land on the issues that still need work. This only changes which tab starts selected - you can switch tabs as usual, and the switch is not saved back to this setting.

- Note: this setting applies on the web. The issue list in the mobile app always starts on All.
- Note: bookmarked issues stay pinned to the top of the list on every tab, but only the ones that match the tab you are on. If you bookmark an issue and then close it, it moves to the top of the Closed tab.

### Canvas settings

Popover and Drawer are the two ways an item (issue / doc / table row) can open on top of the canvas.

#### Popover

A floating window that appears near the item you clicked. The canvas stays visible behind it and you can dismiss it by clicking outside. The popover can be dragged to any position on the screen, resized by dragging either side edge (width only; the width resets to the default when you close it), and can be pinned so it stays open even when you click away (useful for keeping a reference open while working on the canvas).

#### Drawer

A side panel that slides in from the right edge of the screen and pushes nothing off-screen. Stays open until you close it explicitly. Better for longer reading / editing sessions.

#### Issue / Doc / Table row open mode

Choose Popover or Drawer per item type.

#### Canvas background

Plain or Dots (fixed-size dot grid that tracks pan / zoom).

#### Snap to grid; moves and resizes also snap to nearby items

Off or On. Snaps moves and resizes to the grid. Guide lines also appear when the item you are moving or resizing lines up with the edges or centers of nearby items, and it snaps to them. While resizing, only the edges you drag snap. Images resize from their four corners with proportions locked, so an image snaps to one nearby edge at a time, and otherwise snaps to the grid on one axis only. Turn this Off for free placement with no snapping.

#### Hand tool

In menu or Separate. Puts the hand (pan) tool as a dedicated button or hides it inside the tool menu.

#### Tool panel

Menu or Flat. Collapses tools into a menu or shows them as a flat row.

#### Header & tool labels

Minimal (icons only) or Descriptive (icons + text).

#### Note preview style (at low zoom)

Checkerboard, Mosaic, or Skeleton. Controls the placeholder drawn for notes when zoomed far out.

#### Zoom out range

Normal, Wide, or Widest. Sets how far you can zoom out on the canvas. Normal is the default.

- Note: Wide and Widest show more items at once, so a canvas with many items can feel slower to draw. Switch back to Normal if that happens.

#### Show other users' cursors

On or Off.

#### Ambient light

Off, On, or Always. A soft blue glow that breathes across the canvas viewport. Off hides it entirely. On shows it only while the canvas is loading. Always keeps it visible the whole time you are on the canvas. Default On.

#### Minimap

Show or Hide, with position Bottom left or Bottom right.

### Editor settings

Applies to doc and issue editors, and to notes opened in a popover, drawer, or modal. Note text drawn directly on the canvas uses a fixed text size and spacing shared by everyone, so each note keeps the same size and layout for all members; use canvas zoom to adjust how large notes appear. Heading style, heading colors, and emphasis colors apply everywhere, including notes on the canvas.

#### Font size (editor)

Small, Medium, or Large.

#### Spacing (editor)

Compact, Normal, or Relaxed (line height / block spacing).

#### Heading style

`H1, H2, H3...` prefix, markdown `#, ##, ###...` prefix, or None.

#### Heading colors

Set a per-level color for H1-H6. Click the swatch to pick from the color palette, or type a custom hex value (#RRGGBB) in the field. Default inherits the body text color. Each row shows a short sample of heading text in the chosen color, so you can check the color without opening a doc.

If you have never saved an Appearance setting, H1 starts in orange (#EC6742) and H2 in pink (#EC4272). H3-H6 start at Default. The first time you change any other Appearance setting (except Language), including collapsing the minimap, these starting colors are saved as your own heading colors. Choosing Default for a level, or Reset to default, switches to the body text color instead, and the starting colors do not come back. Because they differ from the default, the Reset to default button is shown from the first time you open the modal.

- Note: the starting orange and pink have lower contrast than the body text color. If headings are hard to read, choose Default for that level. A heading color is used in both Light and Dark mode, so a color that reads well in one mode can be hard to read in the other.

#### Emphasis colors

Set the colors used for underlined and bold text. Same swatch and hex input as heading colors. Each row also shows a sample: underlined text on the chosen highlight color, and bold text in the chosen color.

- Underline (U) works like a highlighter pen: it fills the background behind the underlined text. The default is a light yellow. The underline itself is not colored.
- Text sitting on a highlight is always drawn in a dark color, in both Light and Dark mode, like ink on paper. The built-in yellow keeps that text easy to read. If you pick your own highlight color, choose a light one: the text on a highlight stays dark and you cannot change it.
- Bold (B) changes the text color of bold text. Its default adapts to Light, Dark and High contrast.
- Pick your own color to override either default. Your own colors are used as-is in every mode.
- Underline has no Markdown equivalent, so it is not included when content is copied or exported as Markdown. The text itself is kept.

## AI Integration (MCP)

Nproc integrates with AI assistants through the Model Context Protocol (MCP). The integration is designed around Claude, which is the client we have verified it with.

Nproc does not train any AI model on your data and has no built-in AI features. For the limited use of AI tooling by our own administrators in incident investigation and operational monitoring, see the Privacy Policy, section 6.

### Connect Claude

In Claude: Settings > Connectors > Add custom connector and register this URL:

```
https://nproc.net/mcp/v1
```

That connector setup works in Claude on the web, Claude Desktop, and Claude Cowork. In Claude Code, register the same URL from the terminal instead:

```
claude mcp add --transport http nproc https://nproc.net/mcp/v1
```

Claude's permissions through MCP match your own Nproc permissions exactly. It can only access data you have access to and only perform actions you are allowed to perform.

### What Claude can do

#### Read (list / search / get)

- Teams, workspaces, team members, workspace members
- Issues, notes, documents, comments
- Tables (schema + records with filters and sort)
- Document edit history

#### Write

- Create an issue in a workspace
- Create a document in a workspace
- Add a comment to an issue

Documents and issues cannot be edited through MCP once created. To change something you already have, open it in Nproc and edit it there.

Images cannot be added through MCP either. Image markup in content Claude writes is normally not saved, and Nproc says so in the write result. Some markup that Nproc does not read as an image is kept as a plain link instead. To add an image, open the item in the editor.

#### Canvas content

Notes and tables on a workspace are read-only via MCP. Real-time collaborative editing on the canvas is not available through MCP. Issues and docs are readable, and both can be created through MCP, but existing ones cannot be rewritten.

Embedded issue references appear as plain "#123" text when content is read through MCP, and embedded doc references appear as a page symbol followed by the doc title. Writing "#123" or a title in content Claude creates does not link issues or docs; use the editor to embed a reference.

### Common use cases

- Summarize a chat conversation and save it as a doc.
- Take an issue into Claude, analyze it, and post new issues or replies back to Nproc.
- Draft reply candidates from existing reply threads and post them back.
- Summarize a user's issue activity or status.
- Use canvas tables for task tracking: surface assignees' tasks, deadlines, and work at risk.

### Working safely with writes

Reads (list / search / get) are safe to run from anywhere - the worst case is an off-target summary you notice immediately. Writes (create issue, create doc, add comment) create real content in a specific workspace, so a wrong target means cleanup. To keep one simple rule, do your Nproc work - reads and writes - in Claude, with tool confirmations kept on.

The one thing that actually prevents write accidents is pinning the target workspace before any write. Docs and issues are scoped to a workspace; the team is just the container, so the workspace is the target that matters. The discipline is the same every time:

1. Pin the context first. Start the session by naming the workspace, e.g. "Work in workspace Marketing. List it back and confirm before doing anything." Claude echoes the workspace (and its team) so you both agree on the target.
2. Draft before you write. Ask Claude to show the doc/issue body and the destination workspace first, and to write only after you approve.
3. Name the workspace in every write. Always specify the target workspace when you ask Claude to create or update something.
4. Read back to verify. After writing, have Claude get the new doc/issue and report its workspace and title, so you can confirm it landed where you meant.

Claude asks you to confirm tools that change data before it runs them, unless you have turned that confirmation off. Keep it on - do not set Nproc's write tools to run without approval. Pinning the workspace is what makes each confirmation meaningful: it ensures the action you are approving is aimed at the right place.

### Troubleshooting

#### Asked to sign in again

MCP access uses a short-lived authorization that expires for security. When Claude reports it can no longer reach Nproc, reconnect the connector (Settings > Connectors) to sign in again. Your data is not affected.

#### A tool asks for confirmation

Tools that add data (create an issue or document, add a comment) may ask you to confirm before Claude runs them. Read-only tools (list, search, get) do not.

#### Claude cannot see a workspace or issue

MCP respects your Nproc permissions. If you cannot access something in Nproc, Claude cannot either.

## Limits - Quick Reference

| Limit | Value |
| --- | --- |
| Free plan | 1 Admin / 3 Guests / 3 workspaces (archived workspaces included). No Admin or Member invites. |
| Free trial | No trial-specific seat limit. Pro features available. |
| Pro plan | $4 / month per paid user per team, plus applicable tax. At least 1 Admin. Unlimited Members, Guests, and workspaces. Guests limited to 1 workspace / team. Same user in multiple teams = separate seat per team. |
| Copy-paste | Notes and tables can be copied together on the canvas. Issues are not copy-pastable. Moving content into a different workspace requires an active Pro plan for every team involved. This covers the canvas, and images in issue, comment and document bodies - including an image added before you switched workspace. |
| Image upload | 10 MB per image (canvas image import and PDF-rendered pages). |
| PDF import | 100 pages per single upload. Canvas total is not capped - repeat uploads to add more. |
| Invitation email | Sent only to addresses that have no Nproc account. One per address per team. A daily cap applies per Admin; invitations past it still go through without an email. |
| Account data export | 1 request per 24 hours. Download available for 7 days. |