{
  "id": "inspect-eval-framework",
  "type": "entity",
  "name": "Inspect (UK AI Security Institute evaluation framework)",
  "summary": "The open-source evaluation framework from the UK AI Security Institute: composable solvers and scorers, sandboxed execution of untrusted model code, agent primitives, and a registry of 200+ prebuilt evals. The closest thing agent evaluation has to shared infrastructure — METR's Hawk platform builds on it — and the default answer to \"should we write our own harness runner?\" (usually no).",
  "locale": "en",
  "tags": [
    "inspect",
    "harness",
    "tooling",
    "evaluation",
    "open-source"
  ],
  "relations": [
    {
      "rel": "implements",
      "target": "sandboxed-execution"
    },
    {
      "rel": "related",
      "target": "deterministic-vs-probabilistic-evaluation"
    },
    {
      "rel": "related",
      "target": "llm-as-judge"
    }
  ],
  "questions": [
    "What is Inspect and who maintains it?",
    "Should I build my own eval harness or adopt one?"
  ],
  "claims": [
    {
      "id": "c1",
      "text": "Inspect is an open-source LLM evaluation framework by the UK AI Security Institute (GitHub UKGovernmentBEIS/inspect_ai) providing solvers, scorers with both text-based and model-graded options, and over 200 prebuilt evaluations.",
      "sources": [
        "inspect-ai"
      ],
      "confidence": 0.9
    },
    {
      "id": "c2",
      "text": "Inspect runs untrusted model code in sandboxes via Docker, Kubernetes, Modal, Proxmox and other backends through an extension API, and ships agent primitives — ReAct, multi-agent composition, and driving external agents such as Claude Code, Codex CLI and Gemini CLI.",
      "sources": [
        "inspect-ai"
      ],
      "confidence": 0.9
    },
    {
      "id": "c3",
      "text": "METR's Hawk large-scale agent-evaluation platform is built on Inspect infrastructure — third-party adoption at frontier-evaluation scale.",
      "sources": [
        "metr"
      ],
      "confidence": 0.85
    }
  ],
  "takeaways": [],
  "faqs": [],
  "evidence_tier": "secondary",
  "evidence": {
    "level": "industry_observation",
    "source_types": [
      "industry_observation"
    ]
  },
  "moat_flag": false,
  "winning_edge": "Answers the build-vs-adopt question with named capabilities and a named production adopter, where most coverage of Inspect is a feature list without the decision context.",
  "confidence": 0.85,
  "last_verified": "2026-08-08",
  "canonical_url": "https://agentreliability.dev/k/inspect-eval-framework",
  "api_url": "https://agentreliability.dev/api/k/inspect-eval-framework.json",
  "jsonld": {
    "@context": "https://schema.org",
    "name": "Inspect (UK AI Security Institute evaluation framework)",
    "description": "The open-source evaluation framework from the UK AI Security Institute: composable solvers and scorers, sandboxed execution of untrusted model code, agent primitives, and a registry of 200+ prebuilt evals. The closest thing agent evaluation has to shared infrastructure — METR's Hawk platform builds on it — and the default answer to \"should we write our own harness runner?\" (usually no).",
    "url": "https://agentreliability.dev/k/inspect-eval-framework",
    "license": "https://spdx.org/licenses/CC-BY-4.0.html",
    "dateModified": "2026-08-08",
    "citation": [
      {
        "@type": "CreativeWork",
        "name": "Inspect — evaluation framework for large language models",
        "url": "https://inspect.aisi.org.uk/"
      },
      {
        "@type": "CreativeWork",
        "name": "METR — Model Evaluation & Threat Research",
        "url": "https://metr.org/"
      }
    ],
    "author": {
      "@type": "Person",
      "name": "Santiago Santa María Morales",
      "jobTitle": "practitioner — harness engineering and agent evaluation in production"
    },
    "@type": "Article",
    "headline": "Inspect (UK AI Security Institute evaluation framework)"
  }
}