Connect your AI to OpenProject

What your AI can do with OpenProject

Once connected, these tools appear automatically in your AI assistant:

• list_work_packagesread-only

  List work packages, optionally within one project. `statusFilter` open/closed is the usual cut; ids for the other filters come from list_statuses, list_types, list_priorities and list_project_members.

• get_work_packageread-only

  Get one work package. Returns its `lockVersion` — the number an update needs — alongside its fields, with links resolved to {id, name} rather than raw HAL hrefs.

• create_work_packagewrite · needs approval

  Create a work package in a project. `typeId` comes from list_types, `priorityId` from list_priorities, `assigneeId` from list_project_members. Writes to OpenProject, so it goes through the confirmation guardrail when enabled.

• update_work_packagewrite · needs approval

  Update a work package's fields. Only what you pass changes. To add a comment, use add_work_package_comment instead.

• list_work_package_activitiesread-only

  The activity trail of one work package: comments and the field changes that were made, oldest first.

• add_work_package_commentwrite · needs approval

  Add a comment to a work package, changing nothing else. Writes to OpenProject.

• list_work_package_relationsread-only

  Relations of one work package — what it blocks, what blocks it, what it precedes or relates to. The way to answer 'what is holding this up?'.

• create_work_package_relationwrite · needs approval

  Relate two work packages. Writes to OpenProject.

• list_projectsread-only

  List the projects visible to this token, with their id, name and identifier.

• get_projectread-only

  Get one project with its description and status.

• list_project_membersread-only

  Who can be assigned work on a project, with their user id and name. This is how a person's name becomes the id assigneeId expects — without touching the instance's global user directory.

• list_statusesread-only

  Work package statuses with their ids, and which count as closed.

• list_typesread-only

  Work package types with their ids (Task, Milestone, Feature…). A type must be enabled on the target project.

• list_prioritiesread-only

  Priorities with their ids (Low, Normal, High, Immediate).

• list_time_entry_activitiesread-only

  Activities time can be logged against, with their ids. Read them from the time-entry form, which is where OpenProject exposes the values a given work package accepts.

• list_time_entriesread-only

  Logged time, optionally restricted to one project or one work package.

• log_timewrite · needs approval

  Log time against a work package. `activityId` comes from list_time_entry_activities. Writes to OpenProject.

Example prompts

Just ask your AI in plain language — it picks the right OpenProject tool:

• “List work packages, optionally within one project. `statusFilter` open/closed is the usual cut; ids for the other filters come from list_statuses, list_types, list_priorities and list_project_members — from OpenProject, in plain language.”
• “Get one work package. Returns its `lockVersion` — the number an update needs — alongside its fields, with links resolved to {id, name} rather than raw HAL hrefs — from OpenProject, in plain language.”
• “The activity trail of one work package: comments and the field changes that were made, oldest first — from OpenProject, in plain language.”
• “Relations of one work package — what it blocks, what blocks it, what it precedes or relates to. The way to answer 'what is holding this up?' — from OpenProject, in plain language.”

How it works

1. Create your Proxoria account and add your MCP endpoint to Claude or ChatGPT.
2. Connect OpenProject with your API key — each user connects their own account, credentials encrypted at rest.
3. Your AI instantly gets the OpenProject tools. Write actions wait for your approval in the dashboard before anything changes.

Frequently asked questions

Does OpenProject have an official MCP server?

You don't need one. Proxoria exposes OpenProject to your AI through the Model Context Protocol, so you can use it from Claude, ChatGPT, Mistral or any MCP-compatible client — no official OpenProject MCP server required.

What can my AI do with OpenProject through Proxoria?

It gets 17 tools: list_work_packages, get_work_package, create_work_package, update_work_package, list_work_package_activities, add_work_package_comment, list_work_package_relations, create_work_package_relation, list_projects, get_project, list_project_members, list_statuses, list_types, list_priorities, list_time_entry_activities, list_time_entries, log_time. For example, list work packages, optionally within one project. `statusfilter` open/closed is the usual cut; ids for the other filters come from list_statuses, list_types, list_priorities and list_project_members.

Is the OpenProject connector read-only?

No — it can also make changes. Write actions go through Proxoria's confirmation guardrail by default: your AI proposes them, and you approve each one from your dashboard before anything is written to OpenProject.

How do I connect OpenProject to my AI?

Create a Proxoria account, connect OpenProject with your API key, and add your Proxoria MCP endpoint to Claude or ChatGPT. Your AI instantly gains the OpenProject tools.