July 2026 Summaries

ReviewBench is an internally developed benchmark designed to evaluate code review agents by measuring their effectiveness in identifying issues found in real pull requests (PRs) from the LangSmith mono-repo. Unlike synthetic benchmarks, ReviewBench is built on curated comments from trusted reviewers, focusing on concrete, verifiable issues tied to specific codebase standards. The process involves converting raw review comments into tasks using the Harbor format, enabling agents to assess PR changes comprehensively rather than merely scanning for superficial bugs. The benchmark scores agents on coverage and precision, with results indicating that most agents, even with a basic harness, miss many nuanced issues that human reviewers typically catch. However, the study highlights that altering review strategies, such as using structured prompts, can significantly enhance an agent's performance by encouraging a more thorough analysis of code changes. The initiative aims to expand ReviewBench's scope to include a broader array of review tasks, ultimately helping code review agents detect substantive issues without generating excessive review noise.

LangSmith LLM Gateway is now available in public beta, offering a centralized governance layer for managing runtime controls over model calls in production environments to prevent issues like cost overruns and outages. Designed to provide essential controls such as spend caps, rate limits, model fallbacks, and sensitive data handling, the Gateway facilitates consistent governance across agents, models, and providers, thus avoiding vendor lock-in. By allowing engineering teams to define policies once and enforce them uniformly, it mitigates the risks of costly mistakes and outages that can occur when model providers face operational issues. The Gateway supports popular providers and custom models, providing a single interface for routing calls between agents and model providers, and it offers hosted inference for open models through Fireworks to enhance performance and control. Teams have already adopted the LLM Gateway to implement hard spend caps and centralized billing, demonstrating its utility in scaling AI agents effectively. As the Gateway evolves, future enhancements will include broader risk coverage and CI/CD controls to further streamline production model governance.

Deep Agents v0.7 introduces a streamlined base harness, achieving a 65% reduction in base input tokens while maintaining performance, by enhancing context engineering and configurability. This release simplifies token usage by eliminating redundant prompts, trimming tool descriptions, and making TodoListMiddleware optional, which improved efficiency without compromising results. New configurability options allow users to fully customize prompts and middleware, offering control over the harness stack, which was a popular request from users. Validation of these changes was conducted through a comprehensive eval suite across various models, showing consistent performance retention and significant token and cost reductions, particularly for gpt-5.6-luna. Additionally, filesystem optimizations improve core context management by enhancing file interaction capabilities. These updates, alongside deprecated features and new tools, are detailed in the changelog, with the new version available for installation on PyPI and npm.

Liora Korni, a Senior AI Engineer at SimilarWeb, discusses the challenges and methodologies for evaluating agentic systems like Similarweb Data Studio. Unlike traditional software where outputs are consistent, agentic systems produce varied results from the same inputs due to their dynamic nature. Similarweb Data Studio uses LangSmith to evaluate these systems, differentiating between deterministic checks and LLM-as-a-judge scoring to assess outputs. The deterministic checks ensure the correct tools are used, while the LLM-as-a-judge scoring assesses the quality and meaning of the outputs through rubric prompts and feedback. Korni emphasizes that evaluation must be integrated into the product architecture to understand quality changes, avoid miscalibration, and properly attribute scores to specific criteria. Misaligned rubrics can lead to incorrect judgments, as experienced with their Deep Research evaluation, highlighting the importance of aligning evaluations with desired behaviors. By combining golden answers, rubrics, faithfulness checks, and A/B comparisons, SimilarWeb creates a comprehensive evaluation framework that informs product decisions and optimizes agentic system performance.

Over the past year, the data team has overhauled its data stack to better support self-service analysis and agent-based interactions, shifting from a traditional BI tool to a system centered around Hex. This transformation enables agents to handle data requests autonomously by providing business context, clear definitions, and trusted sources, reducing the data team's bottleneck and allowing them to focus on complex queries and strategic projects. The new stack's architecture is designed to accommodate varied user needs, from polished dashboards to conversational interfaces, enhancing accessibility across the company. Contextual details, such as dbt-managed data models, semantic models, workspace guides, and endorsements, are crucial for accurate agent responses, with GitHub and observability tools facilitating ongoing improvements. The shift has led to widespread adoption, with the data agent managing significantly more requests than the data team could previously handle, and has underscored the importance of robust data modeling and context management in fostering reliable data-driven insights.

In the next five years, companies will increasingly integrate AI, either to enhance their operations or as a core component of their products, but relying solely on generic AI models will not suffice for achieving meaningful results. Companies need to "own" their intelligence, which involves controlling critical aspects of AI systems tailored to their specific business needs, rather than building every AI component from scratch. This ownership means managing how AI models are applied, the economic aspects, quality control, and the continuous improvement of AI through learning from interactions. By controlling the model, the orchestration logic, and the contextual data, companies can differentiate their AI applications and ensure they are optimized for their unique requirements. The focus is on creating systems that get better with use, allowing companies to switch models easily, control costs, and ensure the AI system's accountability and auditability. Ultimately, those companies that effectively own and adapt their AI systems will create lasting value and maintain a competitive edge.

Agent design challenges are compounded by the difficulty of evaluation, prompting the development of Deep Agents, an open-source, model-agnostic agent harness. The team behind Deep Agents revamped their evaluation framework, moving from smaller unit tests to comprehensive end-to-end evaluations using Harbor, an open-source framework known for powering Terminal Bench. These evaluations require an environment, instruction, and evaluation script, distinguishing them from simpler LLM evaluations. Three benchmarks—Harbor-Index, 𝜏³-bench, and ContextBench—cover various agent tasks across domains like software engineering and data analysis. The team employs practices such as running tasks multiple times to account for nondeterminism and maintaining a "lite" benchmark for rapid iteration. These benchmarks guide decision-making and iteration, exemplified by preparations for a 0.7 release of Deep Agents, where the team considers removing unnecessary components like the todo-list middleware to optimize performance.

Jensen Huang and Harrison discuss the significance of open agent systems for the future of AI, highlighting NVIDIA's NemoClaw and LangChain Deep Agents as pivotal tools for teams to build and control their agent systems. The Interrupt conference, which previously gathered the agent builder community in San Francisco, is now expanding to New York City and London, offering workshops and talks from industry leaders. LangSmith introduces new features, such as a free trial for Sandboxes, Slack integration, and voice agent tracing, enhancing the development and testing of AI agents. OpenWiki Brains provides a memory layer for agents to utilize various data sources, while Deep Agents and Harbor offer a unified stack for agent evaluation. The Deep Agents Foundation Course is available for those interested in building complex workflows, and upcoming events across multiple cities aim to engage and educate on agent development. Companies like Schneider Electric and Pendo have successfully utilized LangSmith for enterprise-grade AI product development and agent-driven solutions, respectively.

Eval Engineering Skill is a newly launched tool designed to help coding agents build evaluations by utilizing context from a repository and analyzing agent traces. The skill systematically inspects the structure of an agent, identifies patterns from available traces, and proposes abilities to test, while involving user feedback to iteratively approve each evaluation. It generates executable evaluations in Harbor format, which includes a task instruction, an environment setup defined by a Dockerfile, and a verifier to assess task completion. The process begins with mapping the agent's components such as prompts and tools, and understanding data and services that influence its behavior. Users can guide the eval creation by selecting which tools should operate live or be simulated, especially for cost-incurring tasks. The iterative design process allows for refining verifiers based on the agent's trajectory and verifier's reasoning to prevent reward hacking, ensuring evaluations measure the intended capabilities accurately. Containerized evals facilitate rapid experimentation by maintaining stable environments even as agent configurations change, allowing parallel testing and direct comparison of results against fixed targets. This skill is part of the langchain-ai/langchain-skills repository and aims to streamline the building of evaluations, enabling continuous improvement of agent capabilities through reproducible and representative testing environments.

Graph engineering is the latest term to emerge from the realm of AI, joining existing concepts like prompt engineering and loop engineering, aiming to describe the real challenges and design decisions faced in harnessing the power of large language models (LLMs). This approach involves representing agentic systems as graphs, which allows builders to impose structured workflows and control behavior when guiding agents through specific tasks. LangGraph, a framework developed over the past three years, exemplifies this by balancing deterministic and agentic steps, enabling more predictable, powerful, and efficient systems. Unlike traditional deterministic code, nodes in a graph can range from simple LLM calls to full agent runs, allowing for dynamic transitions and flexible runtime variability. While graph engineering is not a novel concept, its recent popularity highlights the evolving strategies in making LLMs more reliable and effective in practical applications, particularly as agents become more capable of handling complex tasks within larger systems.

LangSmith has introduced Python integrations for tracing four popular voice agent frameworks: Pipecat, LiveKit, OpenAI Realtime, and Gemini Live with Google ADK, reflecting the growing practicality and market expansion of voice agents. This development is driven by advancements in voice activity detection, emotive speech models, and large language models (LLMs) capable of real-time dialogue. Like chat-based agents, voice agents require observability to ensure effective operation, necessitating the ability to trace and evaluate interactions within the voice pipeline. LangSmith's new release allows voice and text agents to coexist within the same review and collaboration workflows, offering native support for capturing and tracing voice interactions through architectures like the "sandwich" and the speech-to-speech models. The "sandwich" architecture involves chaining components such as speech-to-text (STT), a text-based agent, and text-to-speech (TTS), while the speech-to-speech architecture utilizes a multi-modal model for direct audio processing. LangSmith's tracing integrations provide full observability into production conversations by capturing metadata, inputs, outputs, latency, and events like interruptions, enabling comprehensive insights into voice interactions.

Apollo is a comprehensive go-to-market platform that aims to streamline the sales cycle from prospecting to analytics by introducing an AI Assistant powered by Deep Agents for goal-based execution. Initially, the platform's extensive range of products and steps overwhelmed users, prompting Apollo to develop a chat-based interface where users can state their goals in natural language, allowing the assistant to manage tasks like prospecting and contact enrichment autonomously. The transition from a supervisor-based architecture to a flexible, skill-driven one with Deep Agents has significantly reduced user confirmation prompts and decreased development time by 80-85%. LangSmith enhances observability and debugging capabilities with a six-layer evaluation strategy called AI Watchtower, ensuring high-quality performance. The AI Assistant has evolved to operate independently of the user interface and can be integrated via API, which has led to rapid adoption and expansion into new areas such as autonomous agents that perform scheduled or continuous tasks. Apollo's future focus includes expanding the assistant's skill library and embedding it as a seamless co-worker in existing user tools.

The LangSmith LLM Gateway serves as a crucial runtime control plane for enterprise AI, converting policy into actionable governance across model interactions, tool calls, and agent processes. As AI agents become integral to production infrastructure, managing their autonomy while ensuring compliance with privacy, security, and AI-specific regulations is vital. Enterprises face increasing complexity in predicting AI spend, maintaining uptime for business-critical agents, and demonstrating consistent policy application. The Gateway enables organizations to authenticate usage, select approved models, enforce data and spending policies, and manage failures while retaining evidence of decisions made. It also provides strategic flexibility, allowing enterprises to adopt more efficient models without re-implementing security measures across applications. The Gateway's integration with tracing, evaluation, and monitoring systems supports continuous improvement, making governance adaptable to evolving models and regulatory landscapes. It ensures that AI governance is not only centralized but also responsive to changes, helping enterprises manage cost, quality, and risk effectively.

LangSmith Engine is designed to identify and rectify issues in other agents by analyzing agent traces, with improvements evaluated using an internal benchmark called IssueBench. IssueBench provides a controlled environment with both clean traces and those containing known issues to determine if the Engine can effectively categorize and group problems for teams to address. It spans 15 tasks across domains like SRE log analysis, software engineering, and customer support, ensuring that the Engine can recognize abstract failure modes rather than just domain-specific patterns. The benchmark uses a fixed set of issue categories to maintain consistency and tests the Engine's ability to produce actionable issue sets, scoring its performance based on correct trace labeling, issue categorization, existing issue assignment, and new issue grouping. This process helps refine the Engine’s capabilities by identifying areas for improvement and ensuring the production of useful engineering work from raw data. The ongoing development of IssueBench aims to enhance its scope and scoring methods to support the growing need for reliable agent evaluation frameworks in production settings.

Financial services leaders are increasingly challenged to demonstrate the return on investment (ROI) of AI initiatives, particularly in complex, multi-agent systems that automate processes like RFP processing and AML compliance monitoring. Traditional financial operations tools struggle to account for the dynamic costs associated with these systems, which involve multiple AI agents interacting with various data sources and APIs. The integration of engineering platforms such as LangChain, LangSmith, and LangGraph with economic intelligence platform Pay-i offers a solution by providing comprehensive observability, cost tracking, and KPI measurement. LangSmith captures detailed traces of agent activities, enabling teams to optimize performance and manage costs effectively, while Pay-i connects these activities to business outcomes by defining and tracking relevant KPIs. This approach allows organizations to quantify time savings and business value, ensuring that AI implementations are not only efficient but also aligned with strategic goals. Financial institutions that effectively leverage these tools can demonstrate clear business value from AI investments, justify expansions, and maintain governance over AI costs and performance.

OpenWiki is an open-source command-line interface tool designed to generate and maintain documentation for codebases by creating a wiki connected to a coding agent, ensuring that the documentation remains current as the codebase evolves. With the release of OpenWiki 0.2, developers can now more easily categorize and structure these wikis, especially for large codebases, by incorporating the OKF format, a proposed standard from Google Cloud, which enhances navigation and reduces time and resources spent by coding agents. The OKF format introduces structured metadata via YAML front matter, which includes fields like title, description, tags, and categories, allowing for improved searchability and organization. It also supports the creation of index.md and logs.md files to summarize content and track updates, respectively, enabling a more efficient review process. Furthermore, OpenWiki's integration with the OKF ecosystem supports compatibility with community-built tools, such as viewers and renderers, and Google has released an OKF wiki visualizer to help users inspect and understand their documentation structure. For new users, OpenWiki can be easily installed and initialized to generate a repository wiki, with more information and installation instructions available on GitHub.

Fleet streamlines the integration of custom agents into Slack, enabling teams to enhance productivity without coding. By allowing one person to build an agent that the entire team can utilize, Fleet agents are tailored to execute repetitive tasks efficiently and in alignment with the company's practices. Unlike generalized chatbots, these specialized agents are designed with specific roles, permissions, and a distinct Slack identity to facilitate seamless collaboration across projects. Users can easily create agents using natural language or prebuilt templates, ensuring the agent is equipped with necessary app connections and knowledge. This integration not only allows for conversational interactions within Slack but also ensures that any additional context or approvals needed are managed within the same thread, preserving project continuity. Fleet provides administrative oversight for controlling access, credentials, and permissions, ensuring agents operate within defined scopes. The recent updates make it easier to deploy these agents in Slack, promoting a cohesive workflow environment.

Agents equipped with their own computing environments are essential for performing tasks autonomously and securely, as traditional text-generating models lack the capability to execute and verify code, akin to a contractor without tools. By enabling agents to run code, access a filesystem, and maintain state across operations, they can independently complete tasks such as debugging, data analysis, and research. However, executing model-generated code raises security concerns, necessitating isolated, hardware-virtualized environments like sandboxes to compartmentalize execution and protect against potential threats. These sandboxes provide agents with the resources to execute tasks safely, observe and control operations, and iterate rapidly while maintaining separation from the host infrastructure. LangSmith Sandboxes, for instance, offer microVM isolation, fast provisioning, and persistent state, allowing agents to perform complex tasks without compromising system security or requiring extensive manual oversight. This approach supports use cases in coding, data analysis, and research by providing secure, scalable, and efficient environments for autonomous execution, thereby enhancing the utility and reliability of agents in various workflows.

LangSmith offers an observability layer that unifies the debugging process across various coding agents, such as Claude Code, Codex, Cursor, GitHub Copilot Chat, Pi, OpenCode, and DeepAgents Code (dcode), by mapping their sessions into a standardized trace schema. This system addresses the challenges of debugging fragmented workflows caused by different agents using diverse structures, metadata, and terminologies. LangSmith allows developers to trace agentic workflows, inspect model calls, and share results with necessary confidentiality, thus enabling them to reconstruct sessions, identify failures, and apply lessons learned to future runs. By providing a consistent framework for analyzing agent behavior, LangSmith facilitates improvements in workflow efficiency, error identification, and cost governance, ensuring that insights from each session enhance overall performance.

OpenWiki Brains is a new framework designed to enhance agents' memory capabilities by enabling them to proactively gather and maintain context from various sources, such as Gmail, Notion, git repositories, Twitter/X, and Hacker News, without explicit instructions. Expanding on the original OpenWiki, which was a CLI tool for codebase documentation, this updated version allows agents to create a general-purpose "brain" that compiles relevant information into a local wiki, ensuring that agents have access to up-to-date context for tasks beyond coding, such as research, planning, and personal workflows. OpenWiki Brains introduces two core concepts: Personal Brain, which focuses on gathering context across connected sources, and Code Brain, which maintains documentation within a git repository. By utilizing connectors to draw information from existing data repositories and employing a proactive approach to memory, OpenWiki Brains aims to provide agents with a durable and automatically updated repository of knowledge, using plain Markdown files for easy navigation and visibility. Future improvements are anticipated in the areas of additional connectors, enhanced retrieval methods, and more sophisticated formats for representing agent memory.

The exploration of harness tuning for open models, specifically using Nemotron 3 Ultra within Deep Agents, highlights the importance of aligning model capabilities with the environment they operate in to maximize performance. While open models like Nemotron 3 Ultra offer cost-effective and adaptable alternatives to frontier models, their effectiveness heavily depends on the compatibility of the harness they are paired with. The study demonstrates that a well-tuned harness, which includes elements such as system prompts, tool descriptions, and middleware, can significantly enhance model performance without altering the model itself. This iterative tuning process involves evaluating model actions, diagnosing behavioral patterns, and making targeted adjustments to improve outcomes. The research underscores that while harness tuning can significantly enhance model performance in specific tasks, it has limitations and cannot compensate for inherent model deficiencies that require post-training solutions. The case study of Nemotron 3 Ultra showed improvements in task performance such as summarization and tool use, indicating that a properly tuned harness can lead to substantial cost savings while maintaining high-quality outputs, though it also highlighted that some long-term behavioral improvements would require changes at the model level rather than through harness adjustments.

The newly announced NemoClaw for LangChain Deep Agents blueprint, developed in collaboration with NVIDIA, aims to enhance the performance of production agents by providing an open and governed framework for enterprises. This blueprint integrates LangChain Deep Agents Code, NVIDIA Nemotron 3 Ultra, and NVIDIA OpenShell runtime to allow teams to customize, secure, and optimize agents for specific workloads, achieving advanced performance at a significantly reduced cost. By tuning the model, harness, evaluations, and runtime together, enterprises can build proprietary systems that reflect their unique domain expertise while maintaining control over the agent stack. The open model layer of Nemotron 3 Ultra is complemented by a tuned agent harness from LangChain and a secure runtime from NVIDIA OpenShell, enabling efficient and transparent deployment of agents. This approach allows companies to fine-tune model weights, customize harnesses, and control runtime based on various requirements, ultimately leading to more efficient and specialized agents. Supported by an ecosystem of partners like EY, Baseten, Fireworks, and Nebius, the blueprint facilitates the deployment of open agentic models in production, ensuring transparency, auditability, and cost-efficiency while meeting enterprise standards.

Deep Agents Code, when run as a NemoClaw blueprint, provides enterprises the ability to utilize coding agents for modernizing legacy systems while maintaining control over sensitive codebases. This approach leverages NVIDIA's open-source technologies, including Nemotron 3 Ultra and OpenShell, to create a secure sandbox environment where coding agents can operate with autonomy but without risk of unauthorized actions. The system is designed to allow enterprises to govern their infrastructure, ensuring code and data residency remain within chosen boundaries. It offers a structured process for modernizing legacy applications by mapping dependencies, refactoring code incrementally, and maintaining an audit trail through all stages of the transformation. This method aims to address the challenges of updating critical systems, such as COBOL-based business logic, by providing a controlled, auditable, and vendor-independent framework for code modernization.

Continual Learning, Harness Engineering, and Post-Training focus on curating data at scale to enhance and improve AI agents by running experiments. This approach was discussed at the AI Engineer World Fair, where the importance of data mining from Traces was highlighted as a crucial tool for companies to understand and improve their agents. Continual Learning involves agents acting in their environment and reintegrating the information gained back into the system, akin to human learning. Traces, which are projections of agent experiences, serve as valuable data to mine for understanding agent behavior. As agents become more complex and produce larger volumes of data, specialized systems like LangSmith Engine have been developed to efficiently process and analyze these traces, finding signals and issues, generating code fixes, and storing crucial information. The integration of open models, which are cost-effective and intelligent, allows for better processing of this data. A practical recipe for agent improvement involves using a combination of Harness Engineering and Fine-Tuning, allowing teams to iteratively enhance agent performance through data collection, evaluation, and continuous experimentation. This iterative process is essential for adapting to increasing data production and enhancing agent capabilities over time.

Schneider Electric is leveraging artificial intelligence to enhance energy efficiency and sustainability across industries, with a focus on electrification, automation, and digitalization. The company operates an extensive AI program through its AI Hub, which involves 350 experts deploying over 60 AI agents to optimize energy consumption, prolong asset lifespans, and boost developer productivity. Central to their strategy is the use of AI to forecast energy demand and production, enabling users to shift electricity usage to cost-effective, eco-friendly times. Schneider's AI operations are underpinned by a robust LLMOps framework built around the LangSmith and LangChain ecosystems, which supports observability, evaluation, and deployment of AI products. This framework ensures data privacy, compliance, and high-quality agent performance, fostering a collaborative environment where subject matter experts can contribute to the development and refinement of AI solutions. Schneider's AI initiatives, such as the internal AI Assistant "One Jo" and the Customer Success Manager Copilot, demonstrate the company's commitment to integrating AI in critical infrastructure while maintaining rigorous cybersecurity standards. Through these efforts, Schneider is advancing its mission to drive sustainable energy management and industrial automation, with a vision of significantly reducing global energy consumption and carbon emissions.

LangSmith addresses the challenges faced by teams using multiple coding agents, such as the increased costs and fragmented data visibility that result from "tokenmaxxing," where excessive spending is mistakenly equated with productivity. The solution involves consolidating data from various agents like Claude Code, Codex, Cursor, and GitHub Copilot Chat into a unified trace model, enabling teams to see and compare expenses across tools in a consistent format. This visibility allows for optimization, where inefficiencies are identified and actionable recommendations are provided by the Engine feature to refine workflows. Additionally, the LLM Gateway offers governance by capping costs at user, team, and organizational levels, and it can integrate open-source models for cost-effective alternatives in appropriate scenarios. By providing a single platform for monitoring, debugging, and measuring coding sessions, LangSmith empowers teams to manage their AI tool usage effectively, ensuring that spending aligns with actual value delivered.

Recursive Language Models (RLMs), introduced by Alex Zhang and MIT CSAIL researchers, aim to counteract the issue of context rot in language models by utilizing programmatic orchestration and dynamic subagents. Unlike traditional models that may struggle with context accumulation over extended sequences, RLMs operate by running code in a REPL (Read-Eval-Print Loop) environment, allowing them to dispatch subagents and recursively process input context. This approach enhances performance by splitting tasks into manageable units and orchestrating them through code rather than relying solely on the model's judgment. Deep Agents, a platform that incorporates RLM support, leverages dynamic subagents and a lightweight code interpreter to facilitate complex workflows across a mix of models, optimizing for tasks like classification and data aggregation. Benchmarking against the OOLONG dataset demonstrates that RLM-enabled agents outperform standard models, particularly in scenarios requiring long-context reasoning, despite higher latency and token costs. By enabling models to write recursive loops for task-specific contexts, RLMs and dynamic subagents provide a structured method to enhance the reliability and scalability of automated workflows.

Pendo's Novus is a product agent designed to automatically detect usability issues in live applications, promptly fix the underlying code, and enhance the overall user experience, achieving a 90%+ success rate in evaluations. The integration of LangSmith tracing has been pivotal, allowing for detailed monitoring and debugging of Novus in production by providing comprehensive trace views that reveal user interactions and system behaviors. Novus utilizes product analytics and session replays to identify and address actionable issues, correlating user behavior with specific code files and suggesting fixes, thereby closing the feedback loop between developers and product managers. This approach allows product teams to maintain high shipping velocity without sacrificing quality, as Novus automatically corrects issues before they become significant problems. The use of LangSmith for trace tagging and cost monitoring ensures efficient resource allocation and insights into customer usage, enabling tailored product improvements and outreach strategies.

OpenWiki is an open-source tool designed to automate the generation and maintenance of documentation for codebases, enhancing coding agents' understanding of repositories. By creating a dynamic wiki linked to a coding agent, OpenWiki ensures documentation remains current as code changes, addressing the challenge of outdated documentation in large repositories with frequent pull requests. Inspired by previous codebase wiki initiatives, OpenWiki integrates with agent instruction files rather than storing extensive documentation within them, allowing agents to access the necessary context efficiently. It supports various model providers and utilizes GitHub Actions to schedule regular updates, ensuring the wiki reflects the latest code changes. OpenWiki aims to streamline the documentation process, reducing the manual effort required from developers and potentially expanding its application to other workflows beyond coding.