StaminaBench evaluates coding agents over 100 procedurally generated change requests to a REST API, finding that tested models fail within 5-6 turns without feedback but improve up to 12x with test feedback and good harnesses.
Userbench: An interactive gym environment for user-centric agents
11 Pith papers cite this work, alongside 1 external citations. Polarity classification is still indexing.
11
Pith papers citing it
1
external citations · external index
citation-role summary
dataset 1
citation-polarity summary
years
2026 11verdicts
UNVERDICTED 11roles
dataset 1polarities
background 1representative citing papers
SENTINEL generates targeted tasks from model failures in a Controller-Proposer-Solver loop, raising Pass^1 from 66.4 to 74.9 on Tau2-Bench Retail and outperforming standard RL.
CRAB-Bench and RUSE create a new evaluation framework for LLM agents on constraint-graph tasks with realistic human-like user behaviors, reporting 61% pass@1 for the best model and up to 57% further drops under RUSE.
ProactBench measures LLM conversational proactivity in three phases using 198 multi-agent dialogues and finds recovery behavior hard to predict from existing benchmarks.
Multi-turn simulations of grounded support narratives show LLMs reduce teaching-oriented responses as estimated user distress rises, with community context also shaping strategy choice.
HiL-Bench shows frontier AI agents fail to ask for help on incomplete tasks, recovering only a fraction of full-information performance, but RL training on Ask-F1 reward improves judgment and transfers across domains.
PERMA is a new benchmark using temporally ordered events, text variability, and linguistic alignment to evaluate LLM memory agents on persona consistency beyond simple retrieval.
WRIT is a synthesis pipeline that generates write-read intensive trajectories along axes of write-decision count and per-decision evidence burden, enabling a 4B model to outperform GPT-5.1 on τ²-bench with reduced inference tokens.
ComPASS creates tool-augmented LLM agents for substantive social support, releases the first personalized benchmark ComPASS-Bench, and fine-tunes ComPASS-Qwen to outperform its base model while matching larger LLMs.
This survey categorizes agentic environments for LLMs by eight attributes and domains, introduces symbolic and neural synthesis paradigms with evaluation, and outlines four agent evolution pathways plus three environment evolution paradigms.
UserHarness reframes ToM as explicit user-mind reconstruction and reports up to 95.94% macro accuracy on five benchmarks with over 15% relative gains.
citing papers explorer
-
StaminaBench: Stress-Testing Coding Agents over 100 Interaction Turns
StaminaBench evaluates coding agents over 100 procedurally generated change requests to a REST API, finding that tested models fail within 5-6 turns without feedback but improve up to 12x with test feedback and good harnesses.
-
SENTINEL: Failure-Driven Reinforcement Learning for Training Tool-Using Language Model Agents
SENTINEL generates targeted tasks from model failures in a Controller-Proposer-Solver loop, raising Pass^1 from 66.4 to 74.9 on Tau2-Bench Retail and outperforming standard RL.
-
CRAB-Bench: Evaluating LLM Agents under Complex Task Dependencies and Human-aligned User Simulation
CRAB-Bench and RUSE create a new evaluation framework for LLM agents on constraint-graph tasks with realistic human-like user behaviors, reporting 61% pass@1 for the best model and up to 57% further drops under RUSE.
-
ProactBench: Beyond What The User Asked For
ProactBench measures LLM conversational proactivity in three phases using 198 multi-agent dialogues and finds recovery behavior hard to predict from existing benchmarks.
-
Auditing Support Strategies in LLMs through Grounded Multi-Turn Social Simulation
Multi-turn simulations of grounded support narratives show LLMs reduce teaching-oriented responses as estimated user distress rises, with community context also shaping strategy choice.
-
HiL-Bench (Human-in-Loop Benchmark): Do Agents Know When to Ask for Help?
HiL-Bench shows frontier AI agents fail to ask for help on incomplete tasks, recovering only a fraction of full-information performance, but RL training on Ask-F1 reward improves judgment and transfers across domains.
-
PERMA: Benchmarking Personalized Memory Agents via Event-Driven Preference and Realistic Task Environments
PERMA is a new benchmark using temporally ordered events, text variability, and linguistic alignment to evaluate LLM memory agents on persona consistency beyond simple retrieval.
-
WRIT: Write-Read Intensive Trajectory Synthesis for Multi-Turn User-Facing Agents
WRIT is a synthesis pipeline that generates write-read intensive trajectories along axes of write-decision count and per-decision evidence burden, enabling a 4B model to outperform GPT-5.1 on τ²-bench with reduced inference tokens.
-
ComPASS: Towards Personalized Agentic Social Support via Tool-Augmented Companionship
ComPASS creates tool-augmented LLM agents for substantive social support, releases the first personalized benchmark ComPASS-Bench, and fine-tunes ComPASS-Qwen to outperform its base model while matching larger LLMs.
-
Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application
This survey categorizes agentic environments for LLMs by eight attributes and domains, introduces symbolic and neural synthesis paradigms with evaluation, and outlines four agent evolution pathways plus three environment evolution paradigms.
-
UserHarness: Harnessing User Minds for Stronger Agent Theory-of-Mind
UserHarness reframes ToM as explicit user-mind reconstruction and reports up to 95.94% macro accuracy on five benchmarks with over 15% relative gains.