{
  "schema_version": 1,
  "builder_plan": {
    "scenario": "Start with one repeated product task that has a measurable failure or cost.",
    "change": "Read the methods and identify one change you can isolate in that workflow.",
    "baseline": "Your current implementation on the same held-out tasks.",
    "steps": [
      "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.",
      "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.",
      "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.",
      "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test."
    ],
    "measure": "Successful outcomes, response time, human rework, and total cost per successful task.",
    "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation."
  },
  "builder_review": null,
  "arxiv_id": "2609.16436",
  "title": "Interpreting and Steering LLM Agents for Social Simulations",
  "summary": "This method replaces pure prompt engineering for controlling agent behavior in simulations with direct manipulation of internal activations: sparse autoencoders decompose representations into interpretable features, and probe-derived steering vectors then push behavior in a target direction. It costs white-box access to the model (ruling out most third-party APIs), an open-weight model that fits the 24 GB GPU, plus the engineering effort to train or obtain SAEs and probes and validate steering per behavior. It can fail when no pretrained SAE exists for the chosen model, when steering vectors generalize poorly across tasks or model versions, when steering degrades unrelated capabilities, and when the claimed advantage over well-designed prompting does not replicate on the reader's workload.",
  "rationale": "The technique is feasible on a single 24 GB GPU with open-weight models and existing SAE releases, but requires internal activations unavailable through APIs and its advantage over prompting is workload-dependent, so small-scale validation is warranted before production adoption.",
  "technique": "SAE- and probe-based activation steering for LLM agents",
  "family": "outro",
  "recommendation": "testar",
  "published": "2026-09-14",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2609.16436",
  "independent_implementations": 0,
  "total_implementations": 0,
  "stars": 0,
  "citations": 0,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "This method replaces pure prompt engineering for controlling agent behavior in simulations with direct manipulation of internal activations: sparse autoencoders decompose representations into interpretable features, and probe-derived steering vectors then push behavior in a target direction. It costs white-box access to the model (ruling out most third-party APIs), an open-weight model that fits the 24 GB GPU, plus the engineering effort to train or obtain SAEs and probes and validate steering per behavior. It can fail when no pretrained SAE exists for the chosen model, when steering vectors generalize poorly across tasks or model versions, when steering degrades unrelated capabilities, and when the claimed advantage over well-designed prompting does not replicate on the reader's workload.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2609.16436",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "SAE- and probe-based steering often outperforms prompt-based manipulation at shifting agent behavior in specified directions, though the advantage depends on the prompting strategy; no quantitative factor is reported.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2609.16436",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "exposure_map": [
    {
      "dimension": "quality",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compute",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "latency",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "operations",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compatibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "security",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "data_and_training",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "reproducibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "risks": [],
  "minimum_test": [],
  "open_questions": [],
  "independent_tests": [],
  "lab_reproductions": [],
  "report_available": false,
  "validation_tier": "unknown",
  "evidence_tier": "unknown",
  "infrastructure_basis": "unknown",
  "equations": [],
  "equations_status": "not_formula",
  "core_kind": "algorithm",
  "equations_fetched_at": "2026-10-14",
  "equations_source_url": "",
  "equations_source_sha256": "",
  "exposures_reviewed_at": "",
  "exposures_source_sha256": ""
}