Factory releases Droid 1.0 coding-agent harness with script mode
Programmatic tool calling is now default; Factory reports 65.2% on Terminal-Bench 4 with Claude Opus 5.5.
1
Stories per day · 30 days
2 RSS
Relatedllm-agents2qwen2agents1benchmark1coding-agent1constrained-decoding1continual-learning1credit-assignment1deepseek1droid1factory1fault-injection1gdpo1inference1kimi1langchain1local-inference1lora1
25 stories, newest first · one card per story
Stories tagged #tool-calling
vLLM constrained decoding lifts Kimi tool-call schema accuracy to 100%
Serving Kimi K2.6 on vLLM 0.28 matched the expected schema only 92% of the time, the authors report.
2
Guide covers running Underdog Saluki 27B for tool calling
The title describes Saluki 27B as a 2-bit Qwen3.8-27B build; the page body was not captured.
1
Larger gpt-oss agents gain no MCP fault tolerance, study finds
MCP-Injector, calibrated on 57,240 threads from 1,774 repositories, cut task scores 60 percent, from 8.08 to 3.23.
WeAct selects LoRA adapters as actions for long-horizon agents
On Qwen3-32B, WeAct beats full fine-tuning by 6.2% and static LoRA merging by 27.5% on eight deep-research benchmarks.
Each Reward to Its Relevant Tokens: Reward-Specific Token-Level Credit Assignment for Multi-Reward RL
MRTPO assigns each reward's advantage only to tokens relevant to that verifier, improving multi-reward RL.
τ-Emotion-Bench: Exposing the Interaction Fragility of LLM Agents through Real-World User Emotions
τ-Emotion-Bench shows emotional users expose reliability gaps in tool-calling LLM agents.
Researchers bypass 6 AI agent frameworks with a simple ‘Repeat-After-Me’ prompt
Researchers used a Repeat-After-Me prompt to make six AI agent frameworks run attacker-written tool calls.
14
ProgAgent: Learning Tool Orchestration through Programmatic Training
ProgAgent trains LLMs to orchestrate tools via code, beating direct tool-calling baselines including at 64k-token states.
Meet the Underdog Saluki 27B: A 2-bit Qwen3.8-27B That Beats the Original at Tool Calling
Saluki 27B, a 7.89 GB 2-bit Qwen3.8-27B GGUF, beats the full model on tool calling.
ConwayResearch/Underdog-Saluki-27B-1.0 — new model trending #30 on Hugging Face
Underdog Saluki 27B 1.0 is a 7.89 GB GGUF of Qwen3.8-27B that retains tool calling for llama.cpp.
POLAR: Ontology-Guided Risk Prevention for Tool-Calling LLM Agents
POLAR prunes risky tool calls using ontology-based reversibility scores, with mixed tau2-bench gains.
OverAct: Measuring and Mitigating Proactive Over-Authorization in LLM Tool-Calling Agents
OverAct finds LLM tool agents routinely access more private data than requests authorize.
Make Sparse Rewards Count: Density-Aware Reward Aggregation for Multi-Reward RL
DARA reweights multi-reward RL by reward density, reaching compliance faster than GDPO on tools and math.
Synthesis Through Simulation: Generating Coherent Enterprise Data via Scalable Agent-System Interaction
SAP's STS generates valid enterprise data by having LLM agents call policy-enforcing simulated APIs.
SLCA-GRPO: Resolving Cross-Segment Credit Misattribution in Tool-Calling RL
SLCA-GRPO assigns separate advantages to tool and summary tokens, improving tool-calling reinforcement learning.
Unreal Agent
Unreal Labs released Unreal Agent, an asynchronous tool harness claiming up to 40% lower costs than Codex.
1
Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents
Growing Harness converts recurring agent control into reusable code, cutting LLM calls by 76 to 92 percent.
EDGEGEN: Improving Tool-Calling Agents Beyond Happy Paths with Synthetic Edge Case Generation
EdgeGen synthesizes database-grounded edge-case tasks that improve tool-calling agents without human annotation.
NemotronLabs VoiceChat: An Open Full-duplex Speech-to-Speech Model with Tool Calling Capabilities
NemotronLabs VoiceChat is an open full-duplex speech-to-speech model that calls tools during real-time conversation.
MTVA-Bench: Evaluating the Language Model Inside Cascaded Voice Agents
MTVA-Bench evaluates language models inside cascaded voice agents across 490 scenarios and 7 languages, exposing 24.4-point score gaps among seven tested models.
Closed-World Resolution Against Tool Hallucination in LLM Agents
Benchmark across ten LLMs documents 322 tool hallucinations and 154 more on MCP, showing model scale does not help and gates cannot reject fabricated calls.
Cactus-Compute/needle3 — new model trending #30 on Hugging Face
Cactus Compute released Needle 3, a 121M-parameter model shipped as an 8–29 MB file for on-device tool calls, extraction, and embeddings.
What Breaks Under Pruning in Smart Homes, and When? Evaluating LLM Degradation Across Architectures and Task Complexity
Pruning study across four LLM architectures finds dense models degrade sharply on smart-home tool calling while MoE models tolerate far more.