{
  "schema_version": 1,
  "builder_plan": {
    "scenario": "A support or data assistant repeats work across related customer requests.",
    "change": "Try the paper's memory change in a replay of past tasks, with separate memory per customer.",
    "baseline": "Your current memory pipeline plus a no-memory control using the same agent.",
    "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": "Task success, stale memories, retrieval misses, response latency, and all memory maintenance costs.",
    "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": "2609.27298",
  "title": "StateComp: Learning When to Compress History in Long Horizon Agents",
  "summary": "StateComp replaces fixed-window or periodic history compression with a learned router that decides, from frozen LM hidden states, when past interactions are safe to compress, grouping READY interactions into spans replaced by summaries. The cost is a two-stage annotation procedure to build KEEP/READY supervision, training an imbalance-aware router, and maintaining a bounded state representation, all adding pipeline complexity beyond a drop-in prompt change. It can fail if the router mislabels still-needed interactions as READY, causing irreversible information loss, and the reported gains come from a single benchmark (WorkBuddyBench), so transfer to other agent workloads is unverified.",
  "rationale": "The approach fits constrained infrastructure since it uses a frozen model and a small router, but the token savings and task-preservation claim require validation on the reader's own agent workload before production adoption.",
  "technique": "State-conditioned history compression for long-horizon agents",
  "family": "memoria_e_contexto",
  "recommendation": "testar",
  "published": "2026-09-23",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2609.27298",
  "independent_implementations": 0,
  "total_implementations": 0,
  "stars": 0,
  "citations": 0,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "StateComp replaces fixed-window or periodic history compression with a learned router that decides, from frozen LM hidden states, when past interactions are safe to compress, grouping READY interactions into spans replaced by summaries. The cost is a two-stage annotation procedure to build KEEP/READY supervision, training an imbalance-aware router, and maintaining a bounded state representation, all adding pipeline complexity beyond a drop-in prompt change. It can fail if the router mislabels still-needed interactions as READY, causing irreversible information loss, and the reported gains come from a single benchmark (WorkBuddyBench), so transfer to other agent workloads is unverified.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2609.27298",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "Reduces total agent and summarization tokens by 52.27% while maintaining task performance on WorkBuddyBench, with 12.67x speedup in representation extraction.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2609.27298",
      "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_fetched",
  "core_kind": "",
  "equations_fetched_at": "",
  "equations_source_url": "",
  "equations_source_sha256": "",
  "exposures_reviewed_at": "",
  "exposures_source_sha256": ""
}