GakuNin RDM
Project ManagementConnect AI agents to GakuNin RDM, the Japanese academic research data management platform. Agents create and update projects, manage contributors, upload and download files, and read registrations so research data stays organized across teams.
What This Integration Enables
GakuNin RDM is the research data management platform operated by Japan's National Institute of Informatics. It is built on the Open Science Framework codebase, so its object model will look familiar to anyone who has used OSF: projects are nodes, nodes nest into components, contributors carry read, write, or admin permission and a bibliographic flag that decides whether they appear in the citation, and a registration is a frozen read-only snapshot of a project at a point in time.
FlowRunner agents work that model directly. They create and update projects, nest components, list and add and remove contributors, list files in a storage provider, stream a FlowRunner file into the project storage, download a file back into FlowRunner storage, and read the registrations created from a project. Authentication is a personal access token scoped to what the researcher granted it. There are no triggers, because the OSF v2 API exposes no webhook subscription endpoints, so flows on this connector are scheduled or driven by an event somewhere else: an instrument finishing a run, a lab notebook entry, a completed analysis job.
The value is not that research data can be moved around faster. It is that the tedious half of good data management, depositing on time, naming consistently, and structuring as you go, stops depending on a graduate student's memory. The half that requires judgment stays with the people accountable for the research, and the platform makes that split explicit rather than hoping for it.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Instrument output deposited on a schedule
A laboratory instrument writes run outputs to a shared location. A scheduled flow collects everything produced since the previous run, calls Get Project to confirm the destination, calls Create Component to open a dated sub-project for the run, and calls Upload File for each output so the bytes are streamed into the project's storage provider. The agent posts the deposited filenames and sizes to the lab channel in Slack. The deposit happens whether or not anyone was thinking about it that week.
Working copies pulled back out for analysis
An analysis pipeline needs the raw files from a project. The agent calls List Files to enumerate the storage provider contents, then calls Download File for the items it needs, which saves them into FlowRunner file storage and returns URLs the rest of the flow can use. The analysis runs, and its outputs are uploaded back into a results component with Upload File. The project retains both the inputs and the derived outputs, linked by the component structure rather than by a naming convention someone has to remember.
Project provisioning and contributor onboarding
A new study is approved. The agent calls Create Project with the title, category, description, and tags drawn from the approved proposal, leaving it private, which is how GakuNin RDM creates projects by default. It calls Create Component for the standard sections the lab uses, and files the approval documents into the project with Upload File after fetching them from Box. It then prepares the contributor list from the proposal and holds it, because adding a contributor grants real access to real data and that is a decision with a name attached.
Human-in-Loop Highlight
Update Project can toggle whether a project is publicly visible. That single field is the difference between a private working space and an open repository, and it moves in a direction that does not come back. Once a project has been public, the metadata may have been indexed and the files may have been downloaded, and setting the flag back to private does not recall any of it. New projects are private by default, which is the right default, and a metadata sync from a lab tracker is exactly the kind of flow that flips it by accident: one field mapped to the wrong column, applied across every project in the tracker.
So the agent never writes the visibility flag from a mapping. It computes the change, calls Get Project to read the current state, and stops with the comparison in front of the principal investigator: "Metadata sync would set Project ab3cd, Perovskite Stability Series, from private to public. It holds 214 files and 3 contributors, and no registration has been created from it. Publish, sync the title and tags only and leave visibility unchanged, or cancel the sync for this project?" The same gate covers Add Contributor and Remove Contributor, because both change who can reach the data rather than what the data says. Everything else, the deposits, the component structure, the file listings, and the registration reads, runs without interrupting anyone. That is the split orchestration as a service is meant to make routine: the agent carries the work, the researcher keeps the decisions that carry consequences.
Agent Capabilities
17 actionsProjects
7- List Projects Lists the top-level projects owned by or shared with the authenticated user, filterable by title text and by tag with page-number pagination. Each entry carries title, category, description, tags, and the public flag.
- Get Project Retrieves full metadata for one project by its five-character node ID, including visibility and timestamps. Used to read current state before proposing any change to it.
- Create Project Creates a new top-level project with a title and optional category, description, tags, and visibility. New projects are private by default, and leaving that default in place is usually the right call for an automated create.
- Update Project Updates project metadata, changing only the fields supplied. It can rename a project, change its category or description, and toggle public visibility, which is the one field on this connector that deserves a person's attention.
- Delete Project Permanently deletes a project. It cannot be undone, requires administrator permission, and a project with components must have those deleted first.
- List Components Lists the child components of a project, the nested nodes used to organize it into logical parts.
- Create Component Creates a child component inside an existing project. Components inherit access from their parent unless changed, which makes them a clean way to structure a project as work arrives.
Contributors
3- List Contributors Lists a project's contributors with their permission level of read, write, or admin, whether they appear in the citation, and their user details.
- Add Contributor Adds an existing user to a project at a chosen permission level, optionally as a bibliographic contributor who appears in the citation. Requires administrator permission, and grants real access to real data.
- Remove Contributor Removes a contributor by user ID, revoking their access to the project. Requires administrator permission.
Files
3- List Files Lists files and folders in a project's storage provider, defaulting to the built-in osfstorage, with each item's name, kind, size, and the file ID used for download.
- Upload File Streams the bytes of a FlowRunner file into a project's storage provider under a given name, returning the created file's metadata. This is the deposit step.
- Download File Downloads a file by its ID into FlowRunner file storage and returns the stored file URL, name, and size. Used to bring project data into an analysis or transformation step.
Registrations
2- List Registrations Lists the registrations created from a project. A registration is a frozen, read-only snapshot of a project and its files at a point in time.
- Get Registration Retrieves one registration with its title, category, registration date, withdrawal status, and the registration schema used.
Users
2- Get Current User Retrieves the profile of the user that owns the personal access token, including user ID, full name, and locale. Used to resolve the acting identity before contributor operations.
- Get User Retrieves the public profile of a user by ID, including full name and any public links. Used to turn a contributor's user ID into a display name.
Frequently Asked Questions
What can FlowRunner do with GakuNin RDM?
FlowRunner agents can run List Projects, Get Project, and Create Project in GakuNin RDM, plus 14 more actions.
Does connecting GakuNin RDM to FlowRunner require OAuth?
No. GakuNin RDM connects to FlowRunner with an API key, no OAuth flow required.
Can GakuNin RDM trigger a FlowRunner workflow automatically?
GakuNin RDM doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with GakuNin RDM
$100 in credits. No card required. Connect in minutes.