Files
2026-07-10 02:19:48 +00:00

77 lines
3.2 KiB
YAML

name: template-comparison
description: Evaluates the dotnet-template-engine/template-comparison skill
type: capability
config:
timeout: 3m
stimuli:
- name: Compare webapi vs webapp side by side
prompt: "I'm deciding between the 'webapi' and 'webapp' templates for a new .NET project. Compare them side by side — show parameters, which supports authentication, AOT, Docker, and controllers, and the available frameworks. Then recommend one for a JSON REST backend."
graders:
- type: exit-success
- type: output-matches
config:
pattern: (?i)webapi
- type: output-matches
config:
pattern: (?i)webapp
- type: output-matches
config:
pattern: (?i)auth
- type: output-matches
config:
pattern: (?i)aot
- type: output-matches
config:
pattern: (?i)docker
- type: output-matches
config:
pattern: (?i)controllers
- type: prompt
rubric:
- The agent used the template-comparison skill to compare the two templates
- The agent inspected each template with `dotnet new <template> --help`
- The agent produced a side-by-side comparison table (not just two separate lists)
- The comparison covered parameters, feature support (auth/AOT/Docker/controllers), and frameworks
- The agent gave a clear recommendation for a JSON REST backend
constraints:
expect_tools:
- skill
- name: Choose between blazor and blazorwasm
prompt: "Which template should I use for an app that needs offline support: 'blazor' (Blazor Web App) or 'blazorwasm' (Blazor WebAssembly Standalone)? Inspect both with `dotnet new <template> --help`, compare the interactivity/hosting trade-offs, and recommend one."
graders:
- type: exit-success
- type: output-matches
config:
pattern: (?i)\bblazor\b
- type: output-matches
config:
pattern: (?i)blazorwasm
- type: prompt
rubric:
- The agent used the template-comparison skill to compare both Blazor templates
- The agent inspected each template's parameters via `dotnet new <template> --help`
- The agent explained the hosting/interactivity trade-offs (server-rendered vs WebAssembly)
- The agent connected the offline-support requirement to a concrete recommendation
constraints:
expect_tools:
- skill
- name: Decide which template fits a background processing scenario
prompt: "I need to run a long-lived background job that processes a queue. Which template should I use — 'worker' or 'console'? Compare them and justify the choice."
graders:
- type: exit-success
- type: output-matches
config:
pattern: (?i)worker
- type: output-matches
config:
pattern: (?i)console
- type: prompt
rubric:
- The agent used the template-comparison skill rather than just creating a project
- The agent compared worker and console templates side by side
- The agent explained why the worker template fits a long-lived queue processor (hosted service lifecycle)
- The agent did NOT create any project files
constraints:
expect_tools:
- skill