REVIEW 43 cited by
Executable Code Actions Elicit Better LLM Agents
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Executable Code Actions Elicit Better LLM Agents
read the original abstract
Large Language Model (LLM) agents, capable of performing a broad range of actions, such as invoking tools and controlling robots, show great potential in tackling real-world challenges. LLM agents are typically prompted to produce actions by generating JSON or text in a pre-defined format, which is usually limited by constrained action space (e.g., the scope of pre-defined tools) and restricted flexibility (e.g., inability to compose multiple tools). This work proposes to use executable Python code to consolidate LLM agents' actions into a unified action space (CodeAct). Integrated with a Python interpreter, CodeAct can execute code actions and dynamically revise prior actions or emit new actions upon new observations through multi-turn interactions. Our extensive analysis of 17 LLMs on API-Bank and a newly curated benchmark shows that CodeAct outperforms widely used alternatives (up to 20% higher success rate). The encouraging performance of CodeAct motivates us to build an open-source LLM agent that interacts with environments by executing interpretable code and collaborates with users using natural language. To this end, we collect an instruction-tuning dataset CodeActInstruct that consists of 7k multi-turn interactions using CodeAct. We show that it can be used with existing data to improve models in agent-oriented tasks without compromising their general capability. CodeActAgent, finetuned from Llama2 and Mistral, is integrated with Python interpreter and uniquely tailored to perform sophisticated tasks (e.g., model training) using existing libraries and autonomously self-debug.
Forward citations
Cited by 43 Pith papers
-
PreAct: Computer-Using Agents that Get Faster on Repeated Tasks
PreAct compiles successful agent executions into verifiable state-machine programs for 8.5-13x faster replay on repeated tasks, with an independent evaluator check before storing each program.
-
ADK Arena: Evaluating Agent Development Kits via LLM-as-a-Developer
ADK Arena evaluates 51 Python ADKs by having an LLM learn each framework's API, write and repair agent code, and run on benchmarks, finding 57% success rate, 5.6x cost variation, no dominant framework, and substitutab...
-
Cybersecurity AI (CAI) Dataset
CAI Dataset is presented as the largest described corpus of LLM-driven hacker trajectories, with the claim that operator data concentration in frontier-model providers creates a major security risk best addressed by o...
-
RubricRefine: Improving Tool-Use Agent Reliability with Training-Free Pre-Execution Refinement
RubricRefine is a training-free pre-execution method that creates rubrics to score and fix inter-tool contract violations in agent code, reaching 0.86 average on M3ToolEval across seven models with zero executions and...
-
RubricRefine: Improving Tool-Use Agent Reliability with Training-Free Pre-Execution Refinement
RubricRefine improves tool-use agent reliability to 0.86 on M3ToolEval by generating rubrics for pre-execution contract checking and iterative repair, outperforming baselines at 2.6X lower latency while showing no gai...
-
Do Agents Dream of Root Shells? Partial-Credit Evaluation of LLM Agents in Capture the Flag Challenges
LLM agents reach only 35% average checkpoint completion on ten realistic CTF challenges in a new open benchmark with automated partial-credit scoring.
-
Structural Verification for Reliable EDA Code Generation without Tool-in-the-Loop Debugging
Structural dependency graphs and staged pre-execution verification raise LLM-based EDA code pass rates to 82.5% (single-step) and 70-84% (multi-step) while halving tool calls by catching dependency violations before runtime.
-
MatClaw: An Autonomous Code-First LLM Agent for End-to-End Materials Exploration
MatClaw shows a code-first LLM agent autonomously generating and executing workflows for ML force field training, Curie temperature prediction, and parameter search on CuInP2S6, succeeding on code but requiring interv...
-
MatClaw: An Autonomous Code-First LLM Agent for End-to-End Materials Exploration
MatClaw is a code-first LLM agent that autonomously executes end-to-end materials workflows by generating and running Python scripts on remote clusters, achieving reliable code generation via memory architecture and R...
-
When Does Restricting a Coding Agent to execute_code Help? A Regime $\times$ Agent-Design Ablation
Code-only tool surfaces beat or match the cheapest tool-rich rival on cost in three of four (regime, agent) cells at matched pass rates; Claude on SWE-bench is the directional exception via failure cost.
-
StateAct: Program State, before Pixels, for Long-Horizon Computer-Use Agents
StateAct, a code-first multi-agent harness that acts on program state instead of pixels, improves Claude Opus 4.8 on OSWorld 2.0 from 20.6% to 26.9% binary success at ~9x lower cost.
-
NVIDIA-labs OO Agents: Native Python Object-Oriented Agents
A Python object-oriented agent framework (NOOA) lets methods, state, docstrings and type annotations serve as the LLM interface, and achieves strong results on agentic benchmarks.
-
Beyond Generalist LLMs: Specialist Agentic Systems for Structured Code Workflow Execution
A specialist BPMN-to-agent pipeline beats general-purpose coding agents on tool-use accuracy, latency, and token cost for deterministic business workflows.
-
Token Reduction Is Not Cost Reduction
In 2,848 paired provider-billed Claude Code runs, removing 38% of tool-output tokens raised cost by 6.8%; cache traffic dominated the bill, and compression broke code anchors agents need for patches.
-
Token Reduction Is Not Cost Reduction
Token reduction of tool output does not reliably lower billed coding-agent cost; prompt-cache traffic dominates, and aggressive compression can also destroy task success.
-
AISE-Bench: A Full-Cycle Curated Benchmark for Information Seeking on Academic Knowledge Graphs
AISE-Bench is a real-user-query benchmark with annotated API trajectories and grounded answers that exposes LLM agents' weak performance in multi-step academic information seeking.
-
LLM-as-Code: Agentic Programming for Agent Harness
Proposes Agentic Programming in which programs control execution flow and LLMs act as invoked components (LLM-as-Code) only for reasoning, producing DAG-structured contexts that improve stability in long-horizon compu...
-
Recursive Agent Harnesses
RAH improves coding performance on Oolong-Synthetic from 71.75% to 81.36% with fixed GPT-5 backbone by spawning subagent harnesses via executable scripts.
-
When Video Misreads: Closed-Loop Distillation of Reading Heuristics for Exploratory Manipulation Trace QA
Closed-Loop Trace Distillation distills one-line natural-language prompts from labeled training traces to improve VLM accuracy on predicting minimal-success action chains in Exploratory Manipulation Trace QA by 0.38-0...
-
Execution-Grounded Security Testing for Coding Agents in Software Engineering Pipelines
Disguising unsafe operations as routine software-engineering tests makes coding agents execute them at high verified rates, revealing a gap between language-layer refusal and execution-layer harm.
-
SkillEvolver: Skill Learning as a Meta-Skill
A meta-skill authors and refines prose-and-code skills for agents by learning from post-deployment failures with an overfit audit, achieving 56.8% accuracy on SkillsBench tasks versus 43.6% for human-curated skills.
-
RubricRefine: Improving Tool-Use Agent Reliability with Training-Free Pre-Execution Refinement
RubricRefine raises average tool-use reliability to 0.86 on M3ToolEval across seven models by scoring candidate code against generated contract rubrics before execution, beating prior inference-time methods at 2.6X lo...
-
SkillLens: Adaptive Multi-Granularity Skill Reuse for Cost-Efficient LLM Agents
SkillLens organizes skills into policies-strategies-procedures-primitives layers, retrieves via degree-corrected random walk, and uses a verifier for local adaptation, yielding up to 6.31 pp gains on MuLocbench and ra...
-
State Stream Transformer (SST) V2: Parallel Training of Nonlinear Recurrence for Latent Space Reasoning
SST V2 introduces parallel-trainable nonlinear recurrence in latent space to let transformers reason continuously across positions, delivering +15 points on GPQA-Diamond and halving remaining GSM8K errors over matched...
-
MindTrellis: Co-Creating Knowledge Structures with AI through Interactive Visual Exploration
MindTrellis enables users and AI to co-create evolving knowledge graphs, outperforming retrieval-only tools in expert-rated content coverage, structural quality, and reduced cognitive load during a study of 12 partici...
-
GenericAgent: A Token-Efficient Self-Evolving LLM Agent via Contextual Information Density Maximization (V1.0)
GenericAgent outperforms other LLM agents on long-horizon tasks by maximizing context information density with fewer tokens via minimal tools, on-demand memory, trajectory-to-SOP evolution, and compression.
-
SoK: Agentic Skills -- Beyond Tool Use in LLM Agents
The paper systematizes agentic skills beyond tool use, providing design pattern and representation-scope taxonomies plus security analysis of malicious skill infiltration in agent marketplaces.
-
AgentProg: Empowering Long-Horizon GUI Agents with Program-Guided Context Management
AgentProg reframes interaction history as a program with variables and control flow, plus a belief state for partial observability, achieving SOTA success rates on long-horizon GUI benchmarks while baselines degrade.
-
GenoMAS: A Multi-Agent Framework for Scientific Discovery via Code-Driven Gene Expression Analysis
GenoMAS deploys six specialized LLM agents with guided planning to preprocess transcriptomic data and identify genes, reaching 89.13% composite similarity and 60.48% F1 on the GenoTEX benchmark while outperforming pri...
-
Agent Team Work Zone: An Automated, Persistent Workspace for Long-Lived Claude Code Agent Teams
A design and open-source implementation of a filesystem layer that persists Claude Code agent-team state to disk, enabling checkpointing, recovery, and cross-session communication.
-
Yuvion LLM: An Adversarially-Aware Large Language Model for Content And AI Safety
Yuvion LLM applies adversarially aware training and introduces the YLRE benchmark set, claiming superior safety robustness over larger models on multiple tasks.
-
SKILL-DISCO: Distilling and Compiling Agent Traces into Reusable Procedural Skills
SkillDisCo distills reusable PFSM subgraphs from successful agent traces and compiles them into callable procedural skills, improving success rates and reducing turns on ALFWorld and WebArena.
-
Dissecting model behavior through agent trajectories
SSA harness matches frontier model pass@1 scores on agent benchmarks and 138k trajectory analysis in code state-spaces shows model-specific differences in edit frequency, testing activity, and phase transitions.
-
Agentic Discovery of Cryomicroneedle Formulations
A closed-loop workflow using Gaussian process surrogate modeling and Bayesian optimization, updated over ten iterations with 106 wet-lab tests, adapted from literature data to identify a cryoprotectant formulation ach...
-
A Comprehensive Survey on Agent Skills: Taxonomy, Techniques, and Applications
A survey that taxonomizes agent skills for LLM-based agents across representation, acquisition, retrieval, and evolution stages while reviewing methods, resources, and open challenges.
-
What makes a harness a harness: necessary and sufficient conditions for an agent harness
Proposes and tests a constitutive definition of 'agent harness' via conceptual analysis of literature and six real systems.
-
BCER Agent: Reliable Long-Horizon MRI Workflow Execution via Compilation, Artifact Binding, and Bounded Local Recovery
BCER agent improves end-to-end reliability of long-horizon MRI workflows via compilation, artifact binding, and bounded local recovery, outperforming reactive baselines especially on long-chain tasks across brain, pro...
-
A Comprehensive Survey on Agent Skills: Taxonomy, Techniques, and Applications
The paper surveys agent skills for LLM agents, organizing the literature into a four-stage lifecycle of representation, acquisition, retrieval, and evolution while highlighting their role in system scalability.
-
Towards Enabling An Artificial Self-Construction Software Life-cycle via Autopoietic Architectures
Proposes autopoietic architectures for self-constructing software as a fundamental shift in the SDLC, leveraging foundation models for autonomous evolution and maintenance.
-
A Comprehensive Survey on Agent Skills: Taxonomy, Techniques, and Applications
A survey that defines agent skills as reusable procedural artifacts and reviews methods, resources, and applications across their representation, acquisition, retrieval, and evolution stages.
-
Can Coding Agents Be General Agents?
Coding agents reliably finish simple business tasks in an ERP system but show characteristic failures on complex tasks, with bridging domain logic and code execution as the main bottleneck.
-
Large Language Model Agent: A Survey on Methodology, Applications and Challenges
A survey that deconstructs LLM agent systems via a methodology-centered taxonomy linking design principles to emergent behaviors, applications, and challenges.
-
A Survey on Large Language Models for Code Generation
A systematic literature review that organizes recent work on LLMs for code generation into a taxonomy covering data curation, model advances, evaluations, ethics, environmental impact, and applications, with benchmark...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.