{ "schema_version": 1, "builder_plan": { "scenario": "An assistant chooses the wrong API or sends arguments your product cannot use.", "change": "Apply the proposed tool-selection or argument-generation change in a sandbox.", "baseline": "Your current agent with the same tool schemas and task budget.", "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": "Correct completed actions, invalid calls, retries, latency, and cost per completed 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": "2609.35760", "title": "TokenCast: Forecasting Token Consumption During LLM Agent Execution", "summary": "TokenCast replaces fixed or naive per-task token budgets with a learned per-segment cost model that composes each execution segment's consumption and context growth, refreshing the forecast at runtime without additional LLM calls (32.8 ms mean cumulative prediction time per run). The cost is building and maintaining a segment-level training corpus of agent traces per agent model and task distribution, plus integrating the predictor into the agent loop's budget-control logic. It can fail when the deployed agent model, tool set, or task distribution drifts from the training traces, when tasks are too few to train a reliable segment model, and the reported gains come from offline replay rather than live production, so real budget-enforcement behavior may differ.", "rationale": "Token-budget control is directly relevant to API-cost management on a constrained budget, but the method requires workload-specific trace collection and per-model training, so its 21.3% saving must be validated on the reader's own agent workloads before adoption.", "technique": "Execution-segment token-cost forecasting for LLM agents", "family": "uso_de_ferramenta", "recommendation": "testar", "published": "2026-09-28", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2609.35760", "independent_implementations": 0, "total_implementations": 1, "stars": 0, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "TokenCast replaces fixed or naive per-task token budgets with a learned per-segment cost model that composes each execution segment's consumption and context growth, refreshing the forecast at runtime without additional LLM calls (32.8 ms mean cumulative prediction time per run). The cost is building and maintaining a segment-level training corpus of agent traces per agent model and task distribution, plus integrating the predictor into the agent loop's budget-control logic. It can fail when the deployed agent model, tool set, or task distribution drifts from the training traces, when tasks are too few to train a reliable segment model, and the reported gains come from offline replay rather than live production, so real budget-enforcement behavior may differ.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.35760", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "In offline budget-control replay, TokenCast uses 21.3% fewer tokens on average than a fixed-budget policy at matched trace completion (factor ~1.27); mean absolute error of token forecasts is reduced by an average of 14.5% versus the strongest comparator across 96 combinations of 4 task suites and 6 agent models.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.35760", "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": "" }