Newest papers

Structure Before Intelligence: Gating What Agentic Scientific Workflows Release

Ashish Acharya, Aman Al Amin, Hong Qin · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.19507 · cited by 0 works (OpenAlex) · open-access PDF

French

Un flux de travail scientifique cible peut se terminer et encore libérer des résultats erronés: une erreur numérique silencieuse, une entrée stable ou un défaut dans une étape en amont partagée peut contaminer plusieurs agents de sortie tandis qu'une branche sœur reste correcte. Nous traitons la confiance dans ces résultats comme une décision de libération. Un harness autour du graphique acyclique dirigé du flux de travail (DAG) décide sur quelles sorties peuvent être utilisées: il applique des contrôles obligatoires, contient des défaillances à la branche affectée, recompte des versions de base suivies, ajoute des validateurs et libère un résultat seulement si un modèle de risque estime sa probabilité d'être invalide comme faible. Les agents de modèle de langue

Machine translation (Meta M2M100); the English original follows.

Abstract

An agentic scientific workflow can run to completion and still release wrong results: a silent numerical error, a stale input, or a defect in a shared upstream step can contaminate several outputs while a sibling branch stays correct. We treat trust in such results as a release decision. A harness around the workflow's directed acyclic graph (DAG) decides which outputs may be used: it enforces mandatory checks, contains failures to the affected branch, recomputes from tracked versions, adds validators, and releases a result only if a risk model estimates its probability of being invalid as low. Language-model agents work inside the harness, choosing corrective actions and reviewing releases, but cannot override it. On 60 held-out paired instances of two workflows with injected faults, we switched components on and off. Each supplied a different property. Containment and recovery raised the number of correct results. Validation supplied safety: fixed rules without validation released 37 invalid results, while validating every candidate released none. Risk estimation supplied efficiency, matching that safety with 31% fewer validators at somewhat lower yield, although better targeting than random validation was not shown. Lineage evidence helped on one workflow. The language model added no measurable benefit: whether invalid results escaped depended on the harness, not the model. This proof of concept points to a practical path: placing agents inside explicit release-control structure can keep released scientific results trustworthy, and it gives more capable agents a measurable baseline to improve on.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Workflow, Directed acyclic graph, Computer science, Matching (statistics), Task (project management), Work (physics)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange