Context
The team discussed how patterns could become more actionable through AI skill files — structured prompt files that help agents identify and apply relevant patterns based on context. Naveen has been experimenting with this approach and proposed establishing a GSF-hosted community repository for green software skill files, bootstrapped with a small number of seed skills to demonstrate the concept.
The Green AI Committee has been identified as the right resource to inform the initial seed skill files, given their expertise in AI sustainability practices. UCL's work on sustainability annotations for architecture diagrams was also cited as a related reference point.
The intent is that skill files would be contributed by the community over time — one per SDLC lifecycle category as a starting structure — and that users could combine them based on their context (e.g. a skill file for requirements analysis, one for architecture review, one for prompt engineering).
What needs to happen
Acceptance criteria
- A repository exists with a clear contribution standard and at least 2–3 seed skill files
- Seed skill files demonstrably reference and apply GSF patterns
- Community members can contribute skill files following published guidelines
- At least one end-to-end example is documented showing an agent consuming a skill file to assess a real artefact (e.g. an architecture document) against patterns
Dependencies
- Green AI Committee availability and engagement
- Patterns catalogue content stability (skill files should reference a reasonably settled catalogue)
Context
The team discussed how patterns could become more actionable through AI skill files — structured prompt files that help agents identify and apply relevant patterns based on context. Naveen has been experimenting with this approach and proposed establishing a GSF-hosted community repository for green software skill files, bootstrapped with a small number of seed skills to demonstrate the concept.
The Green AI Committee has been identified as the right resource to inform the initial seed skill files, given their expertise in AI sustainability practices. UCL's work on sustainability annotations for architecture diagrams was also cited as a related reference point.
The intent is that skill files would be contributed by the community over time — one per SDLC lifecycle category as a starting structure — and that users could combine them based on their context (e.g. a skill file for requirements analysis, one for architecture review, one for prompt engineering).
What needs to happen
Acceptance criteria
Dependencies