Data · dataset · 2026
Data package for "How developers split roles among them and a coding agent and what 'collaboration' means"
Listed in ZivaHub and Deakin Research Online and DMU Figshare — shown once because both records carry DOI 10.6084/m9.figshare.34031112.v2
<p dir="ltr">This is the data package accompanying an article with the following abstract:</p><p dir="ltr">
Description
Background
Coding agents such as Claude Code become more and more competent, so that their role in the development process can become more and more prominent.<br>Objective: How do developers characterize the role-split between them and their agent? How would they ideally want to collaborate with their agent?<br>
Read the rest (5 more)
Methods
We perform a 3-step, mixed-methods study. We apply qualitative open coding to 2025 early adopters’ blog articles for deriving the role-split they used. We perform a quantitative survey of 2026 practitioners for measuring their general attitudes and role-split.</p><p dir="ltr">We develop a method for coding-agent-supported open coding of survey freetext responses and apply it to a question regarding the desired manner of collaboration between human and agent.<br>Results: The 2025 practitioners split roles such that all intent formation, all major design decisions, and all accept/non-accept decisions were reserved for the human; the agent would work almost exclusively in supporting roles.
Accountability was completely and firmly human. In 2026, the split and accountability attitude are fundamentally the same, even for most respondents not actually using coding agents. Gradually, however, the split is now different, with much more agent involvement mixed into many work types that were previously human-only.
The 2026 developers’ notion of collaboration has three facets: Predominantly a separation into human-only, agent-only, and collaborative work steps, with defined conditions and goals of the collaboration. Second, sometimes, accommodating</p><p dir="ltr">certain personal preferences without an engineering justification. Third, sometimes, a holistic view of collaboration covering the entire development process.<br>Conclusions: The trend towards more agent involvement is likely to continue, so that defining one’s role relative to the agent will need to be a recurring activity.
The different views on collaboration are likely a source of friction and conflict for teams and future work should find out how to cope with that.</p><p><br></p><p dir="ltr">The data package contains questionnaire, raw data, open coding skill, evaluation software, some computed results, and a PDF of the article draft itself.</p>
Links
Where it is published
- DOI doi.org/10.6084/m9.figshare.34031112.v2 ↗
DOI / persistent id · from zivahub uct ac za
Catalogue records · 1
- OAI-PMH record api.figshare.com/v2/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Af… ↗
metadata API · from zivahub uct ac za
Topics
- From keywords
- Computer Science & AI · Computer Science & AI · Computer Science & AI · Earth & Environmental Science · Earth & Environmental Science · Earth & Environmental Science · Empirical software engineering · Empirical software engineering · Empirical software engineering · Engineering · Engineering · Engineering
Provenance · 3 source records, 17 field assertions
| Source | Key | Last seen | Raw |
|---|---|---|---|
| ZivaHub | oai:figshare.com:article/34031112 | 5 d ago | JSON v1 |
| Deakin Research Online | oai:figshare.com:article/34031112 | 5 d ago | JSON v1 |
| DMU Figshare | oai:figshare.com:article/34031112 | 5 d ago | JSON v1 |
| Field | Assertion | Extractor | Evidence |
|---|---|---|---|
| access_level | source · zivahub uct ac za | connector:zivahub_uct_ac_za@1.0.0 | |
| concepts[field].anzsrc:field:461202 | mapping · figshare dmu ac uk | vocabulary-mapper@1.0.0 | keywords['Empirical software engineering'] |
| concepts[field].anzsrc:field:461202 | mapping · zivahub uct ac za | vocabulary-mapper@1.0.0 | keywords['Empirical software engineering'] |
| concepts[field].anzsrc:field:461202 | mapping · dro deakin edu au | vocabulary-mapper@1.0.0 | keywords['Empirical software engineering'] |
| concepts[field].local:field:computer-science-ai | mapping · zivahub uct ac za | connector:zivahub_uct_ac_za@1.0.0 | |
| concepts[field].local:field:computer-science-ai | mapping · figshare dmu ac uk | connector:figshare_dmu_ac_uk@1.0.0 | |
| concepts[field].local:field:computer-science-ai | mapping · dro deakin edu au | connector:dro_deakin_edu_au@1.0.0 | |
| concepts[field].local:field:earth-environmental | mapping · figshare dmu ac uk | connector:figshare_dmu_ac_uk@1.0.0 | |
| concepts[field].local:field:earth-environmental | mapping · zivahub uct ac za | connector:zivahub_uct_ac_za@1.0.0 | |
| concepts[field].local:field:earth-environmental | mapping · dro deakin edu au | connector:dro_deakin_edu_au@1.0.0 | |
| concepts[field].local:field:engineering | mapping · dro deakin edu au | connector:dro_deakin_edu_au@1.0.0 | |
| concepts[field].local:field:engineering | mapping · zivahub uct ac za | connector:zivahub_uct_ac_za@1.0.0 | |
| concepts[field].local:field:engineering | mapping · figshare dmu ac uk | connector:figshare_dmu_ac_uk@1.0.0 | |
| description | source · zivahub uct ac za | connector:zivahub_uct_ac_za@1.0.0 | /metadata/dc/description |
| license | source · zivahub uct ac za | connector:zivahub_uct_ac_za@1.0.0 | /metadata/dc/rights |
| publication_date | source · zivahub uct ac za | connector:zivahub_uct_ac_za@1.0.0 | |
| title | source · zivahub uct ac za | connector:zivahub_uct_ac_za@1.0.0 | /metadata/dc/title |