Workshop proceedings
Proceedings of the 1st Workshop on Reliable Tool-Using Agents
RTUA 2026
- 12 June 2026 – 13 June 2026
- Lisbon, Portugal
- Meridian University and Open Systems Lab
Sample volume. The event, organisers and papers are fictional demonstration content.
About this volume
RTUA brings together researchers studying why tool-using language agents fail and how to make them dependable. The workshop covered failure analysis, sandboxing, logging and measurement of reliability in real deployments.
RTUA-2026-proceedings.pdf
Papers (5)
Failure analysis
- pp. 1–9560
Full paperBest paper
Retry Budgets for Tool-Using Agents: When Is Trying Again Worth It?Aisha Rahman, Lena Hartwell
10.xxxx/ctm.pr.2026.0001
downloads - pp. 19–24190
Short paper
Detecting Silent Tool Failures from Output DistributionsCarlos Medina, Kofi Mensah
10.xxxx/ctm.pr.2026.0003
downloads
Security and oversight
- pp. 10–18430
Full paper
Sandboxing Agent Tool Calls with Capability TokensMarcus Tan, Jonas Berg
10.xxxx/ctm.pr.2026.0002
downloads - pp. 25–30340
Short paper
A Minimal Logging Schema for Reconstructable Agent RunsSofia Lindqvist, Julia Fernandes
10.xxxx/ctm.pr.2026.0004
downloads
Measurement
- pp. 31–33150
Extended abstract
Perturbation Suites for Measuring Agent Reliability: Lessons from Three DeploymentsHana Kovács, Aisha Rahman
10.xxxx/ctm.pr.2026.0005
downloads