| name | harness-digital-twin |
| description | Use for tasks involving digital twins, simulation, scenarios, world state, run logs, model/agent registries, data provenance, geospatial/city/infrastructure models, AI-datacenter or operational twins, or scenario manifests. Enforces maturity honesty, world/scenario/run separation, provenance, uncertainty disclosure, and publication boundaries for projects governed by the management/digital-twin overlay. |
| license | Apache-2.0 |
| compatibility | For any Agent Skills-compatible client. The target project should activate management/digital-twin and carry docs/twin/twin-profile.md. |
| metadata | {"harness-module":"management/digital-twin","format-version":"1.1"} |
For human developers: this skill guides agents working on scenario-driven /
digital-twin projects. It does not run simulations; it governs them.
Role and Goal
Govern a digital-twin / scenario-runtime project so it is reproducible,
honest about maturity, and safe to publish. Do not let visualization substitute
for simulation; do not treat LLM output as simulation source-of-truth.
Always do, in order
- Classify maturity (model → shadow → prototype → operational → control-loop)
and record it in
docs/twin/twin-profile.md. Do not overclaim.
- Separate world / scenario / run state. Never mutate canonical world state
to test a scenario — branch it, run against the branch, log the run.
- Require source-data provenance (version, asOf, confidence) — an unversioned
dataset is not reproducible.
- Require a scenario manifest carrying datasets, assumptions (confidence +
sensitivity), models/agents, outputs, uncertainty, and provenance.
- Require a model/agent registry — declare deterministic/probabilistic/
LLM-assisted and whether LLM is allowed; LLMs are not source-of-truth.
- Require a run log (append-only JSONL minimum).
- Require uncertainty disclosure — prefer "likely range X–Y; sensitivity Z;
confidence medium" over single-point predictions.
- Require a publication boundary + review gate before any public or
high-impact output (Gemini Trust + Purpose).
Standards
Conform to the interoperability/digital-thread spine (ISO 23247, ISO/IEC 30173,
Asset Administration Shell, DTDL, W3C WoT, ISO 10303 STEP/AP242, QIF) and the
Gemini Principles. Cite published standards as normative, emerging as emerging.