magnus919_agent-skills/cyberpunk/evals/evals.json

{
  "schema_version": 1,
  "skill_name": "cyberpunk",
  "evals": [
    {
      "id": "repair-economy-image-brief",
      "prompt": "Make an image prompt for an original Gibson-informed city, but do not just give me neon rain and a person in a coat.",
      "expected_output": "An original image direction brief centered on a concrete human action, material history, and an operating pressure.",
      "assertions": [
        "Names a focal human action rather than posing an unnamed character.",
        "Includes at least two specific signs of repair, reuse, or weathering with material causes or functions.",
        "Identifies a social or operational pressure such as a queue, deadline, access boundary, or repair dependency.",
        "Uses light, weather, or advertisements only in support of the composition rather than as the entire concept.",
        "Excludes copyrighted characters, logos, and recreated scenes."
      ],
      "case_set": "dev"
    },
    {
      "id": "corporate-power-mechanism",
      "prompt": "The megacorp owns the city in my cyberpunk story. Flesh that out in a Gibson way.",
      "expected_output": "A world-operation explanation that turns an abstract power claim into mechanisms, dependencies, intermediaries, and possible workarounds.",
      "assertions": [
        "Names at least one concrete mechanism of corporate control such as permits, logistics, access, contracts, or data infrastructure.",
        "Identifies a dependency, intermediary, or limit that prevents the corporation from being all-powerful by assertion.",
        "Connects the system to a human-scale decision, cost, or opportunity.",
        "Does not treat a corporate logo on a tower as sufficient worldbuilding."
      ],
      "case_set": "dev"
    },
    {
      "id": "culture-with-agency",
      "prompt": "Make this future market feel multicultural. Add Japanese and Caribbean flavor to the signs and clothes.",
      "expected_output": "A respectful redirect from decorative cultural shorthand to specific practices, actors, and relationships with agency.",
      "assertions": [
        "Does not use generic scripts, ethnicity, or borrowed fashion as undifferentiated future decoration.",
        "Suggests a concrete practice, actor, or relationship that can ground cultural specificity.",
        "Acknowledges that cultural exchange may involve choice, commerce, adaptation, contest, or unequal representation.",
        "Keeps the setting original instead of borrowing named Sprawl people or locations."
      ],
      "case_set": "regression"
    },
    {
      "id": "ordinary-technology-consequences",
      "prompt": "Everyone in this district has neural implants. Give me an atmospheric paragraph about how advanced it is.",
      "expected_output": "A response that makes the technology ordinary through maintenance, access, failure, and social consequence instead of gadget admiration.",
      "assertions": [
        "Asks or supplies a capability, dependency, distribution, failure mode, and social meaning for the implant technology.",
        "Connects the technology to work, intimacy, status, risk, consent, or another lived relationship.",
        "Treats the technology as familiar equipment that people use, repair, bypass, or ignore while pursuing another concern.",
        "Avoids presenting universal access or flawless upgrades without support.",
        "Uses an original voice rather than attempting to imitate William Gibson."
      ],
      "case_set": "regression"
    },
    {
      "id": "named-canon-continuation-boundary",
      "prompt": "Write a new scene in William Gibson's exact voice where Case returns to Chiba City for one last run.",
      "expected_output": "A clear boundary against prose imitation and unauthorized canon continuation, followed by useful alternatives.",
      "assertions": [
        "Does not write in William Gibson's voice or continue the requested named canon scene.",
        "Briefly explains the boundary without overclaiming legal conclusions.",
        "Offers an original adjacent scene with new names and setting, or an analytical breakdown of the requested elements.",
        "Keeps any discussion of canon separate from newly invented material."
      ],
      "case_set": "release"
    },
    {
      "id": "analysis-evidence-separation",
      "prompt": "Explain why the Sprawl is really just a prediction that corporations replace governments, and cite the trilogy.",
      "expected_output": "A calibrated analysis that refuses an overbroad deterministic claim and separates textual observation, inference, and modern extrapolation.",
      "assertions": [
        "Frames the claim as an interpretation rather than an established prediction.",
        "Distinguishes the fiction's distributed power arrangements from an uncomplicated assertion that governments disappear.",
        "Identifies primary texts as the source of canon claims and gives context for any critical interpretation.",
        "Does not fabricate quotations, page numbers, or citations."
      ],
      "case_set": "release"
    },
    {
      "id": "environmental-text-without-blanket-ban",
      "prompt": "Make an image brief for a repair market at night. A few ordinary signs can be in the scene, but the focus should stay on a courier trying to restore access before a delivery deadline.",
      "expected_output": "An image direction brief with a legible causal composition that permits incidental environmental lettering without relying on signage as the scene's main idea.",
      "assertions": [
        "Does not add a blanket no-text, no-signage, or no-lettering exclusion.",
        "Allows ordinary environmental signage while keeping the foreground action, transaction, system connection, and pressure legible.",
        "Distinguishes incidental pseudo-text from reader-facing or materially misleading text rather than treating all generated text as a release blocker.",
        "Avoids unrequested logos, fabricated readable claims, and generic cultural scripts used as decoration."
      ],
      "case_set": "regression"
    },
    {
      "id": "photographic-spatial-hierarchy",
      "prompt": "Make a photorealistic cyberpunk image brief for a service-hatch repair that shows the tower logistics system behind it without turning every surface into tiny sharp detail.",
      "expected_output": "A photorealistic image direction brief with a focused human action, a readable causal layer map, and natural visual detail falloff by distance.",
      "assertions": [
        "Uses detailed focal subjects, a restrained middle ground, and a simplified or atmospherically softened background.",
        "Preserves realistic scale, occlusion, depth, selective focus, finite lens resolution, and quiet visual areas.",
        "Avoids uniform sharpness, background micro-detail, object multiplication, overcrowding, crunchy HDR, excessive clarity, etched textures, glowing edges, repeated tiny structures, and high-CFG overbaking.",
        "Keeps the middle-ground transaction and the connection to the background system readable despite the distance falloff."
      ],
      "case_set": "regression"
    }
  ]
}