Glossary · Objects and system context
Harness (scaffolding)
Everything around the model that turns it into an agent: prompts, tools, memory, retrieval, skills, orchestration, permissions, and stopping logic.
Also called: scaffolding
Defined in
Handbook · ClaimPopulation, environment, harness, and validity domainEvery score was earned in some world. Declare the population, environment and harness, and say where the claim stops.Related terms
- Eval harness
- The machinery that administers an evaluation and records trials. It is part of the instrument, distinct from the agent harness, which is part of the system under test.
Where it comes up
- Evaluation is measurementHandbook · Foundations
- Constructs, facets, and nuisancesHandbook · Claim
- Exposure, modification, and accessHandbook · Regime
- Task validity and outcome validityHandbook · Validation
- Harness architecture and reproducibilityHandbook · Lifecycle
- Evaluating agentsHandbook · Agents and self-improvement
- Self-improvement and evaluator co-evolutionHandbook · Agents and self-improvement
- Guarantee before you estimateHandbook · Practice
- Notation and formula cardHandbook · Reference