Skip to content

The workflow: entry points and phases

Data Director follows the process flow in the Blueprint (§8): six phases, joined at one of three entry points. This page is a map of that flow, so you can see where you are in your work and what the Data Director can help with from there.

The workflow is yours. The Data Director assists, recommends and flags; it does not direct or control your work, and you can skip, repeat or reorder steps to suit your project.

Choose an entry point

Researchers do not always start at the beginning. Say which of the three describes you.

Entry Where you are You join at
Entry 1 Starting a new project, before data collection begins Phase 1
Entry 2 Data already collected, being prepared for sharing Phase 2
Entry 3 Ready to deposit data into a repository Phase 3

All three continue through every later phase to the end.

In the chat, you can answer with "Entry 2", or describe your situation in your own words ("I have collected my data and want to share it"). On a first message that names no entry point, the Data Director asks which one you mean. It then lists the steps from that entry onwards and the agents that can help with each.

The map

flowchart TD
    E1(["Entry 1<br/>New project"]) --> P1
    E2(["Entry 2<br/>Data collected"]) --> P2
    E3(["Entry 3<br/>Ready to deposit"]) --> P3
    P1["Phase 1<br/>Pre-collection setup"] --> P2
    P2["Phase 2<br/>Governance and constraints"] --> P3
    P3["Phase 3<br/>Repository selection"] --> P4
    P4["Phase 4<br/>Data preparation and<br/>metadata creation"] --> P5
    P5["Phase 5<br/>Deposit and submit"] --> P6
    P6["Phase 6<br/>Post-publication"]

The phases and their steps

Each step names the Blueprint requirement it applies (R1 to R12, from §7.1), where there is one. Steps marked optional may be left out.

Phase 1: Pre-collection setup

Entry 1 starts here.

  • Draft the data documentation (R5)
  • Select controlled vocabularies and ontologies (R3)

Phase 2: Governance and constraints

Entry 2 starts here. These checks come before repository selection and constrain everything after them.

  • Check funder, journal and institutional data sharing policies
  • Identify sensitive data
  • Handle sensitive data
  • Confirm whether an embargo is required
  • Verify data sharing commitments against the Data Management Plan (R8)

Phase 3: Repository selection

Entry 3 starts here. Choose the repository before you create metadata: it decides the metadata schema, vocabularies and validation rules for Phase 4.

  • Check whether the funder, journal or institution mandates a repository
  • Recommend a repository where none is mandated (R1)
  • Confirm or arrange persistent identifier (PID) assignment (R7)
  • Check access options and deposit costs
  • Verify repository certification (optional)

Phase 4: Data preparation and metadata creation

  • Clean and version the data files
  • Recommend open formats (R3)
  • Create or review data documentation: READMEs and data dictionaries (R5)
  • Identify the appropriate metadata standard
  • Complete metadata fields with FAIR-aligned, discipline-appropriate vocabularies and ontologies (R2, R6)

Phase 5: Deposit and submit

Metadata is validated before submission, while you can still act on what is found.

  • Validate metadata quality (R4)
  • Assign a licence
  • Link to related research outputs: publication DOIs, ORCID, ROR and grant IDs (R7)
  • Set access restrictions
  • Generate a private peer review link (optional)
  • Upload the files
  • Review and finalise the submission

Phase 6: Post-publication

Publication is not the end. This phase is a continuing loop.

  • Record the deposit in institutional systems where required
  • Monitor access, downloads and citations
  • Check embargo release
  • Evaluate the FAIRness of the published objects and discover related datasets (R12)
  • Retrofit FAIR-aligned metadata for existing datasets (R9)
  • Link datasets to related research outputs (R11)

In every phase, each action an agent takes on your data and metadata is recorded with its evidence: see What the Data Director records.

How you interact with it

  1. Say where you are. Start a conversation and name your entry point.
  2. Read the plan. The Data Director lists the steps from your entry point. For each step it shows the requirement and the agents that serve it.
  3. Hand a step to an agent. The Data Director delegates the step to an agent through the harness, or you open the agent directly and use its Run tab (Running an agent). The harness checks the agent kept to the contract and records the run.
  4. Check the result. Every result carries an outcome and the evidence it rests on. You decide what to do with it.
  5. Move on. Continue to the next step, or return to an earlier one.

Steps with no agent

An agent joins a step by declaring the steps it serves. Where no registered agent serves a step, the Data Director says "no agent yet", and gives the requirement if the Blueprint names one. It does not skip the step or do the work itself. The set of steps with an agent depends on which agents your deployment has registered, so the list differs between deployments.

The Blueprint phases panel in the app shows the same phases and, for each step, the agents here that serve it. It does not yet show where your own project has reached in them (TODO(#160)).

What is not covered

  • Data Management Plans. Creating, reviewing and updating a plan is out of scope for the initial phase, so the steps for it in the Blueprint's figures are not offered (TODO(#284)).
  • Governance as an entry point. Routing researchers down open, restricted or unshared paths is deferred by the Blueprint to v2.0.
  • The figures themselves. The steps above are taken from the figure captions in §8.2. They may omit steps or decisions that the figures show (TODO(#245)).

Where this is defined

The entry points and phases are data in the SDK (dd_sdk.process_flow), the same for every agent and deployment. Orchestration records the decision; the Blueprint chapter is the source.