Claude connector (MCP)

List the people on this team with their user IDs. Call this to turn a name mentioned by the user into an assigneeId or ownerId before filtering or creating anything.

No inputs.

Report which Fishladder account this connection is authenticated as, which team it is reading from, and what that account can do there. Use it when they ask who they are connected as, and whenever results look wrong or a team name cannot be found — the common cause is that the connector was authorized by a different account than they have in mind, and no other tool can reveal that.

No inputs.

List the Fishladder teams this person belongs to, and say which one this connection reads from. Use this when an answer looks wrong or empty and they expected data — the connection covers one team at a time, and they may be thinking of another. Also use it when they ask which teams they have, or before switch_team if the name they gave is ambiguous. To change team, call switch_team — it takes effect immediately, with no reconnect. If the team they want is not in this list, call whoami: the connector is probably authorized as a different account than they expect.

No inputs.

Change which Fishladder team this conversation reads from and writes to. Accepts the team name as the person would say it. Use list_teams first if you are unsure what exists, or if the name given is ambiguous. Everything after this — tasks, CRM, wiki — comes from the new team, and their role and modules there may differ.

List tracker projects (task boards). Call this first when the user names a board or you need a projectId for search_tasks or create_task.

Create a tracker board — the columns and statuses that tasks live on. Use this when someone is starting a new piece of work and there is no board for it yet; check list_tracker_projects first so you do not make a second board for something that already has one. A board is created from a template, which decides its starting columns and statuses. Pass skipTasks true unless they want the template's example tasks as well.

Search tracker tasks by free text, project, assignee, or priority. Use this for any question about work items — what someone is working on, what is overdue, or whether something already exists before creating it.

Create a tracker task. Search first with search_tasks to avoid duplicates, and resolve the assignee with list_team_members rather than guessing an ID.

Update a tracker task: change its status, mark it complete or reopen it, or edit its title, description, assignee, priority or due date. Find the task with search_tasks first. Only the fields you pass are changed; everything else is left alone. To move a task to a named column, read the status IDs from search_tasks rather than guessing.

Attach a file to a tracker task — PNG, JPEG, GIF, WEBP or PDF — passed as base64. Use it for something you produced in this conversation, such as a chart you generated or a report you assembled. You CANNOT use it for a screenshot or image the person uploaded to the chat: you were shown that image, you do not hold its bytes, and inventing base64 would attach garbage. In that case say so plainly and give them the task link so they can drop it on the task themselves. Maximum 10MB.

Return a link that opens a task with its attachment area ready, so the person can paste or drop a file in one step. Use this whenever they want to attach something YOU do not hold the contents of — above all a screenshot or file they pasted into this chat, which you were shown but cannot forward. Say plainly that you cannot upload it for them, give them this link, and tell them to press Cmd-V (Ctrl-V on Windows) when it opens.

Add a Google Drive, Docs, Sheets or Slides link to a task's Google Drive section. Use this — not the description — when someone wants a Drive file attached to a task. Adding the same file twice is harmless. Find the task with search_tasks.

List the Google Drive files linked to a task (its Google Drive section).

List roadmap projects (the strategic plan: title, status, start/target dates). Use this when the user asks what is planned, in flight, or shipping — not for individual tasks.

Create an item on the roadmap — a strategic piece of work with a status and optional start and target dates. This is the plan, not the day-to-day board: use create_task for the work underneath it. Search list_roadmap_projects first so you extend the plan rather than duplicating something already on it.

Update a roadmap item: change its status, dates, description or title, or link it to a tracker board. Find it with list_roadmap_projects first. Only the fields you pass are changed; everything else is left alone.

List what is due or scheduled in a time frame, across the whole workspace: task due dates, roadmap project target dates, deal expected-close dates, quote expirations, interviews, security questionnaire due dates, audit milestones and calendar events. Use this for any "what is due / coming up / deadlines / what did we miss" question, optionally narrowed to named projects. Finished work is left out unless asked for.

List the spaces (sections) of the team's River wiki in Fishladder. A space can sit inside another space: parentSpaceId names it, and path reads like "Customer Success / Cursor". Call this before create_river_page to choose where a new page belongs.

Search the team's River wiki inside Fishladder — internal docs, specs, runbooks, process and policy pages written by this team. Use this for questions about how the team does something ("how does our escalation process work", "what is our release process"). This is distinct from a general file store such as Google Drive: it searches only pages authored in Fishladder. Call get_river_page for the full text of a result.

Fetch the full text of one River wiki page in Fishladder, as Markdown, along with its version number and any open comment or suggestion threads. Use after search_river_pages when the excerpt is not enough to answer the question. The result can be edited and sent back through update_river_page (pass version as baseVersion) without losing formatting.

Attach a comment to a specific passage of a River wiki page WITHOUT changing the page — the passage is highlighted and the comment appears in a margin panel, labeled as coming from Claude. Use this for review feedback instead of editing reviewer notes into the body. Quote the passage exactly as get_river_page returned it.

Propose replacing a specific passage of a River wiki page with new text. This does NOT change the page: the proposal appears as a tracked change that a person accepts or rejects in Fishladder. Use it for concrete wording fixes; use add_page_comment for open questions. Quote the passage exactly as get_river_page returned it. An empty replacement proposes deletion.

Reply to an existing comment or suggestion thread on a River wiki page. Thread IDs come from get_river_page (openThreads) or from add_page_comment/suggest_page_edit results.

Mark a comment thread on a River wiki page as resolved — the highlight is removed and the thread moves to the resolved list. Only for plain comments: a suggestion is accepted or rejected by a person in Fishladder, never through a tool.

Create a new page in the team's River wiki inside Fishladder — a doc, spec, runbook or process note. Search with search_river_pages first so you update an existing page instead of duplicating it. If the team has more than one space, call list_river_spaces and ask which one to use.

Edit an existing page in the team's River wiki inside Fishladder. Only the fields you pass change — sending just a title renames the page and leaves its body alone. Passing content replaces the whole body rather than appending to it, so read the page with get_river_page first and send everything it should end up saying, along with the version you read as baseVersion — if the page changed meanwhile, this call fails and you should read it again. Inline comments survive the rewrite as long as their quoted text does. The previous version is kept in the page history. To move the page, pass spaceId and/or parentId (or moveToTopLevel) — a move needs no content, and child pages move with it. The result includes the page's location (space and parent pages).

Link a River wiki page to something else in Fishladder — most often a task (so the spec shows up on the task) or a board, roadmap project, CRM record, quote, job or candidate. Use it after create_river_page when the page belongs to a piece of work. Get the page ID from search_river_pages / create_river_page and the task ID from search_tasks.

List the interactive diagrams and documents hosted on the team's Fishladder. Returns titles and IDs, not the documents themselves — call get_canvas for one of those. Check here before creating a canvas, so an existing diagram gets updated rather than duplicated.

No inputs.

Fetch a hosted diagram including its full HTML source. Read this before calling update_canvas so an edit builds on what is there rather than replacing it blind.

Publish an interactive HTML diagram or document to Fishladder and get back a URL the team can open. Use this for diagrams whose interactivity carries the meaning — hover detail, expandable nodes, filtering. For a static flowchart, a Mermaid code block in a River wiki page is lighter and easier to edit later. Call list_canvases first to avoid creating a second copy of a diagram that already exists.

Update a hosted diagram. Only the fields you pass change — sending just a title renames it and leaves the document alone. Passing html replaces the whole document rather than appending to it, so call get_canvas first and send everything it should end up with, not a fragment. The previous document is kept in the canvas history.

List the team's slide decks in Fishladder with their build status. A deck still showing "building" is not finished — decks are generated asynchronously and take minutes.

No inputs.

Fetch one slide deck with its slides and live build status. Call this to check whether a deck you started has finished: status goes building → generating_images → assembling → complete, or failed with a reason. Read it before editing a slide, so an edit builds on what is there.

Start building a PowerPoint deck from the team's Fishladder data — roadmap, projects, goals and feedback, chosen with the include flags. IMPORTANT: this returns immediately with a deck that is NOT built yet. Generation is asynchronous and takes minutes. Tell the person it has started, then use get_presentation to check whether it finished — do not report it as ready. Limits: 10 builds an hour per team, 2 at a time. For an interactive diagram, use create_canvas instead — slides are PowerPoint and cannot hold one, though a slide can link to a canvas.

Add a slide to an existing deck. Appended at the end unless you give sortOrder. Slides are PowerPoint and therefore static — for an interactive diagram, publish a canvas with create_canvas and reference its URL from a bullet.

Edit one slide. Only the fields you pass change — sending just a title leaves the body alone. Passing content replaces the whole body rather than merging into it, so call get_presentation first and send everything the slide should end up with.

Search CRM companies by name or domain. Use this when the user mentions a customer or prospect and you need their record or companyId.

Search CRM contacts by name, email or job title. Use this to answer "who do we know at X", and before create_crm_contact so you update an existing person instead of duplicating them.

Add a company to the CRM. Search with list_crm_companies first — a duplicate customer record splits their deals and history across two entries, which someone then has to merge.

Add a person to the CRM. Search with list_crm_contacts first to avoid duplicating them. Attach them to a company with companyId, resolved via list_crm_companies — a contact with no company is hard to find again.

Update a company already in the CRM — its lifecycle stage, industry, domain, website or description. Only the fields you pass change. Find the company with list_crm_companies.

Update a person already in the CRM — their job title, email, phone, company or lifecycle stage. Only the fields you pass change. Find them with list_crm_contacts.

List CRM deals with value, stage and status. Use this for pipeline questions — what is open, what closed, how much is in flight. Values are returned in cents.

Search the team's connected HubSpot portal for companies or contacts by name, domain or email. Use this to find the HubSpot record ID before link_hubspot_record. Returns name, domain or email, and a link to the record in HubSpot.

List the HubSpot companies and contacts linked to a task, roadmap project or River page, with a link to each record in HubSpot.

Link a HubSpot company or contact to a task, roadmap project or River page. Find the record with search_hubspot first. Linking something already linked is a no-op. Nothing is changed in HubSpot.

Remove a HubSpot company or contact link from a task, roadmap project or River page. Get the link ID from list_hubspot_links. The record itself is untouched in HubSpot.

Search customer feedback entries. Use this when the user asks what customers are asking for, or wants evidence behind a roadmap decision.