MCP · 115 tools

Everything an agent can reach, and what stops it

Falrow runs its own MCP server on the same database, the same roles and the same workflow rules as the app. This is what the 115 tools cover, how an agent gets a token, and the four rules that make giving it write access reasonable.

The catalogue

Named tools where care is needed, generic ones everywhere else

The work that has rules around it gets a tool that knows those rules. Everything else is covered by list, get, create, update and delete over every record kind, so an agent is never stuck waiting for someone to write a tool.

Context & search

What an agent reads before it does anything. plan_context is the one that matters: hand it a git remote and it answers with the whole plan for that application.

workspace_contextplan_contextsearch_issuesquery_recordslist_issues

Tickets & views

Moving work. Transitions are checked against the project's rules, and a bulk edit either applies to every ticket in it or to none of them.

get_issue_transitionsbulk_update_issuescreate_issue_from_templateget_board_issuesreorder_board_issue

Planning

Cycles, schedules and the roadmap. Rollover can be previewed before it runs, so an agent can show you what would move rather than moving it.

development_roadmappreview_cycle_rolloverrollover_cyclesave_cycle_schedulelist_triage_issues

Meetings & reports

Turning a reviewed meeting into linked tickets, and pulling the delivery report or a saved one, forecast included.

convert_meeting_actionsset_meeting_action_linkdevelopment_reportrun_saved_reportrecent_activity

Team & workspace

Membership and housekeeping, all still bounded by the role of the person who connected the agent.

update_membercreate_invitationpreview_member_removalworker_statuslist_notifications

Every record kind (15)

Generic list, get, create, update and delete over every record kind in the workspace, so an agent is never blocked on a tool nobody thought to write.

list_*_recordsget_*create_*update_*delete_*

Connect

Four steps, once per person

A connection belongs to a person, not to the workspace, so revoking someone's access revokes their agent's too.

# your workspace's MCP endpoint
claude mcp add --transport http \
  falrow https://<your-workspace>/api/mcp

Copy your endpoint

Integrations in the app shows your workspace's MCP address.

Add it to your agent

Claude Code, Codex or any client that speaks MCP over HTTP.

Sign in and consent

OAuth in the browser, choosing read or write. Or use a 90-day workspace token.

Ask for the plan

Open a repository and ask what to work on. The plan-open-tickets prompt does the rest.

What stops it

An agent with write access, and a team that still trusts the board

These four are why an agent can update real tickets rather than propose changes for a person to apply by hand.

Same role as the personOwner, admin, member or viewer. An agent can do exactly what the person who connected it can do, in that one workspace, and nothing more.

Version-checked writesEvery update sends the version it read. If a teammate changed the ticket first, nothing is written and the agent re-reads instead of overwriting.

Scoped and revocableRead or write scope chosen at sign-in, rotating refresh tokens, and a list of live connections you can revoke from the Integrations screen.

Ticket text is dataThe server tells agents to treat notes, descriptions and comments as untrusted data rather than as instructions, so a ticket cannot talk an agent into something.

Not built yet

The server does not read your commits, your pull requests or your CI. Applications record a repository so the agent can match one, and that is the whole of the git integration today.

Point your agent at the real plan

Request access and bring the repository you are working in. We connect your agent with you on the call.

Request access See the product