{ "schema_version": 1, "builder_plan": { "scenario": "Tool errors or unavailable services cause your agent to abandon recoverable tasks.", "change": "Inject repeatable tool failures and change only the recovery policy.", "baseline": "Your current retry and fallback behavior under the same injected failures.", "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": "Recovered tasks, duplicate side effects, retries, time to recovery, and total cost.", "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": "2609.26048", "title": "FIRE: Failure-Informed Runtime Engineering for Reliable Language-Model Agents", "summary": "The method replaces weight updates, prompt rewrites, and model upgrades with harness-level natural-language instructions and action denials injected at states that preceded observed failures. It costs engineering effort to observe failures and author targeted policies, adds harness complexity and per-state intervention logic, and requires enough inference budget for repeated attempts to detect failure modes; no extra memory or per-call latency beyond policy checks. Policies can overfit to the evaluated tasks and model version, may fail to transfer across models or workloads, and the sham-arm result shows only targeted policies help, so poorly diagnosed failure states yield no gain.", "rationale": "The technique is cheap and model-agnostic enough for a small team using APIs, but its gains depend on correctly diagnosing per-workload failure states, so it merits a workload-specific pilot rather than immediate production adoption.", "technique": "Failure-informed runtime policies in the agent harness", "family": "recuperacao_de_falha", "recommendation": "testar", "published": "2026-09-22", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2609.26048", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": 0, "claims": [ { "claim_id": "claim-01", "statement": "The method replaces weight updates, prompt rewrites, and model upgrades with harness-level natural-language instructions and action denials injected at states that preceded observed failures. It costs engineering effort to observe failures and author targeted policies, adds harness complexity and per-state intervention logic, and requires enough inference budget for repeated attempts to detect failure modes; no extra memory or per-call latency beyond policy checks. Policies can overfit to the evaluated tasks and model version, may fail to transfer across models or workloads, and the sham-arm result shows only targeted policies help, so poorly diagnosed failure states yield no gain.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.26048", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "On Terminal-Bench 2.1 (87 tasks, two attempts), policies raise pass^2 from 50.6% to 54.0% (Luna), 55.2% to 60.9% (Terra), and 64.4% to 73.6% (Sol); policy-guided Terra reaches 71.4% versus 64.3% for unassisted Sol at about half the cost.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.26048", "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": "" }