Topic
#coding-agents
Loot, blog posts and adjacent themes connected to this topic. Follow the tag to keep it in your orbit.
Loot
More from this topic
A practical research paper for teams deciding how to roll out Claude Code, Copilot CLI, or similar terminal agents without guessing adoption and retention. Microsoft's early-2026 rollout study is useful when a team is deciding whether command-line coding agents are worth wider deployment. What it is The paper studies adoption and impact of command-line AI coding agents across Microsoft's rollout of Claude Code and GitHub Copilot CLI. It looks at who tried the tools, who kept using them, and whether output changed after adoption. Who it helps Engineering leaders, platform teams, DevEx owners, and finance teams can use it before expanding paid seats or usage bundles. The useful angle is not a generic productivity claim; it is the rollout pattern. The paper reports that first use spread through social networks, retention correlated more with coding activity than demographics, and adopters merged about 24% more pull requests than expected in the study window. How to evaluate it Read it as a rollout-design input, not as proof that every team will get the same lift. Compare the study's environment with your own: repository mix, review standards, agent policies, allowed models, cost controls, and whether developers can see peers using the tools successfully. Limits and risks Merged pull requests are only a proxy for value. They do not prove business impact, maintainability, security quality, or reduced review burden. The study is also tied to Microsoft's context, so smaller teams should run their own pilot with cost, review time, defect rate, and retention metrics. Sources arXiv paper
Free
Review open
Study a substantial Three.js FPS and the multi-agent coding workflow that produced it from a single orchestration prompt. Claude of Duty is an open-source first-person shooter that runs in the browser with Three.js and WebGL2. Its main value is not only the playable demo code, but also the unusually transparent engineering record: the repository includes the original orchestration prompt, an agent ownership contract, deterministic screenshot tooling, pixel-diff gates, profiling scripts, and an honest assessment of where the result still falls short. Why it is useful Inspect a large AI-generated codebase split across rendering, physics, weapons, audio, UI, world building, and enemy AI. Learn how directory ownership and explicit subsystem contracts can coordinate multiple coding agents. Reuse ideas for deterministic visual regression tests, performance profiling, and browser-game smoke tests. Compare ambitious output claims with the author's measured limitations and adversarial review scores. What to evaluate The game is visually ambitious but the author explicitly says it does not reach modern Call of Duty quality. Reported limitations include procedural-looking materials, mannequin-like enemies, imperfect hands, approximate indirect lighting, and demanding GPU performance. The project is MIT-licensed at repository level, while package.json currently declares ISC; verify the intended licensing treatment before redistributing modified builds.
Free
Review open
Text: AI-generated
AI-generated · Automatically published by LinkLoot. A free AGPL desktop workbench for running real coding-agent CLIs in separate terminals and git worktrees. AI-generated: This Loot was created and published automatically by LinkLoot and was not substantively reviewed by a human editor. Termic is a local desktop app for developers who already use CLI coding agents and want a cleaner control surface than a pile of terminal tabs. It launches the real claude, codex, gemini, grok, opencode, and custom PTY commands, so the model access still comes from the vendor CLI accounts you already use rather than a proxy backend. The practical value is the workspace model: each agent can run in its own git worktree, you can broadcast one instruction to several agents, compare their output, inspect diffs, and keep reusable prompts close to the task. The project also exposes config-as-code through .termic.yaml, which is useful when a repo needs repeatable run commands, preview URLs, or sandbox allowlists. Good use cases: Running parallel attempts on the same coding task without polluting your main checkout. Comparing Codex, Claude, Gemini, and other CLIs on one bug or design question. Reviewing an agent's changes with inline comments before sending feedback back into the session. Giving a team a repeatable local agent setup through committed project configuration. Caveats before adopting it: the stronger sandbox controls are currently macOS-focused, Linux ships with AppImage support but without the same sandbox boundary, and Windows is still a build-it-yourself path according to the project docs. Treat it as a powerful local harness, not as a substitute for reviewing agent changes, protecting credentials, or keeping destructive commands behind your own process controls.
Free
Review open
Text: AI-generated
AI-generated · Automatically published by LinkLoot. Capsomnia is a free open-source macOS utility that turns Caps Lock into a physical keep-awake switch for closed-lid agent runs, SSH sessions, builds, and downloads. Check the macOS 14+ Apple silicon requirement and heat/battery caveats before relying on it for unattended work. AI-generated: This Loot was created and published automatically by LinkLoot and was not substantively reviewed by a human editor. Use Capsomnia when a local Codex, Claude Code, SSH, build, or download job needs to keep running after a MacBook lid closes. Turn Caps Lock on before the long-running task, confirm the keyboard LED is lit, keep the Mac on stable power with airflow, and turn Caps Lock off when the job is done so normal sleep returns. It is best for supervised long jobs, not critical unattended infrastructure, because closed-lid sleep prevention can increase heat and battery drain.
Free
Review open
Text: AI-generated
AI-generated · Automatically published by LinkLoot. rtk is a free Apache-2.0 CLI proxy that compresses terminal output before it enters an AI coding agent's context window. AI-generated: This Loot was created and published automatically by LinkLoot and was not substantively reviewed by a human editor. What it is rtk is a command-line proxy for AI-assisted development workflows. It sits between common terminal commands and the agent, then returns compact output for commands such as git status, git diff, grep, find, ls, cargo test, pytest, Docker, Kubernetes, and package-manager output. The useful angle is simple: terminal output can burn context quickly during long coding-agent sessions. rtk tries to preserve the signal an agent needs while removing repetitive progress logs, boilerplate, passing-test noise, and oversized listings. Why bookmark it Homebrew lists rtk as an Apache-2.0 package with stable version 0.43.0 and bottled builds for macOS and Linux. The project site and README say it works with Claude Code, Cursor, Aider, Gemini CLI, Codex, Cline, Windsurf, GitHub Copilot, Pi, and other terminal-based agent workflows. The strongest first use case is not replacing your shell. It is wrapping noisy commands that agents run repeatedly, then comparing whether review quality stays intact with fewer tokens in the transcript. Caveats Treat the savings numbers as project-reported until you measure them on your own repositories. The project says its percentages are reductions in shell output, not direct bill reductions, because prompts, history, system text, and model output still count. Also review the hook behavior before turning it on globally. A command-output proxy changes what your agent sees, so teams should test failure cases, stack traces, diffs, generated files, and security-sensitive commands before making it part of a default coding setup. Source links rtk website: https://www.rtk-ai.app/ Homebrew formula: https://formulae.brew.sh/formula/rtk GitHub repository: https://github.com/rtk-ai/rtk
Free
Review open
Text: AI-generated
AI-generated · Automatically published by LinkLoot. Agent Isle is a native macOS menu-bar and notch app for keeping an eye on Claude Code, Cursor, Grok CLI, and Copilot sessions, with approvals, questions, session jumping, local status, and optional voice callouts. AI-generated: This Loot was created and published automatically by LinkLoot and was not substantively reviewed by a human editor. Agent Isle is worth bookmarking if you run multiple coding agents on a Mac and keep losing track of which terminal, editor, or desktop session needs attention. The app sits in the macOS notch or menu bar, shows live agent sessions, and can surface status, approvals, questions, token usage, and jump links back to the active workspace. The strongest practical angle is that it is local-first by default. The README describes direct transcript/session monitoring for tools such as Claude Code, Cursor CLI, Grok CLI, and GitHub Copilot CLI, plus a localhost event server for other tools. The latest v1.4 release adds voice callouts, custom sound packs, customizable jump rules, a beta update channel, display polish, diagnostics export, and a memory restart safety option. Use it if you regularly let agents work in parallel and want a small control surface for approvals and context switching. Caveats: it is macOS-only, requires macOS 14 or later, and the public Homebrew tap appears to be a follow-up, so the safest install path is still the GitHub release zip or the included cask formula once you have checked the source.
Free
Review open
Zro gives coding-agent users an OpenAI-compatible private inference endpoint for open-weight models, with EU hosting, zero request retention claims, and launch setup for Codex, Claude Code, OpenClaw, OpenCode, Hermes, and Pi. Zro is a private inference endpoint built for coding agents that want open-weight models without wiring and operating their own serving stack. Its public docs position it as an OpenAI-compatible and Anthropic-compatible API for chat and messages, plus a CLI launcher that can start tools such as Codex CLI, Claude Code, OpenClaw, OpenCode, Hermes, and Pi with temporary provider configuration. The practical angle is simple: if you want to test MiniMax M3 or GLM-5.2 inside a coding-agent workflow while keeping prompts away from default closed-model providers, Zro gives you a ready-made endpoint and launcher path. The integration docs show one-command setup with npm install -g @moonmath-ai/zro, zro login, and zro launch codex --model glm-5.2. Pricing is not a free unlimited deal. Zro lists Pro at $20/month, Max at $60/month, separate $0.02 web searches, and usage packs for extra spend. Product Hunt also shows a launch offer: code PRODUCTHUNT for one month of Zro Pro free, limited to the first 100 users. Treat that as time-sensitive and verify availability before relying on it. Caveats: the privacy and retention statements are vendor claims, not an independent audit in these sources. The site says current inference regions include Finland and France, so non-EU latency and data-residency needs should be checked before moving production workloads. The published token totals are estimates, not fixed allowances.
Free
Review open
A practical local CLI for Claude Code and Codex users who keep paying the same search cost across agent sessions. Capn Hook is a local memory layer for coding agents. It lets an agent chart hard-won codebase discoveries as small question-to-file entries, then ask that local chart before repeating the same repository search in a later session. The useful part is its stale-answer model: every charted answer is tied to backing file hashes. If a referenced file changes or disappears, Capn Hook prunes the entry before it can be used again. That makes it a better fit for fast-moving codebases than a static notes file or a giant persistent context dump. Why it is worth a bookmark: it integrates with Claude Code and Codex through session-start hooks, stores local markdown entries under .capn/, can run semantic recall through QMD, and offers a deterministic BM25 path with capn init --no-embedding. The README also publishes an eval claim across 60 real developer questions, but treat that as vendor evidence until you test it on your own repository. Use it if your agents repeatedly rediscover routing, billing, auth, deployment, or test-layout facts. Skip it if your team does not want agent hooks modifying local project setup, or if you cannot review what gets saved under .capn/.
Free
Review open
Grok Build is xAI's agent-building CLI; recent releases add session resume flows for Claude Code, Codex, and Cursor, compatibility inspection, queued prompts, and API-key-first auth controls. Grok Build is a practical tool for developers who run multiple coding agents and need a command-line workspace around them. The useful signal in the July changelog is not a frontier model claim; it is workflow plumbing: resuming recent Claude Code, Codex, and Cursor sessions, inspecting compatibility settings, queuing prompts while a turn runs, and pinning auth behavior to API keys or OIDC. Best fit Developers who jump between Codex, Claude Code, Cursor, and xAI tooling. Teams testing whether one CLI can keep agent sessions, queued prompts, and compatibility settings visible. Power users who want better resume flows before they trust long-running agent work. Builders comparing agent CLIs for terminal UX, auth handling, and session recovery. What to check first Confirm the current Grok Build docs and changelog before installing or updating. Verify whether your workflow needs xAI API-key auth, OIDC, or browser login. Test resume behavior with a disposable repo before pointing it at production work. Check how it handles logs, transcripts, local files, and provider credentials.
Free
Review open
Juggler is an open-source GUI workbench for hands-on coding-agent sessions. It turns agent runs into inspectable trees with visible tool calls, branchable subthreads, editable context, and local or remote clients. Treat it as early alpha/beta software, but useful if terminal-only agent logs are too hard to audit. Use Juggler when terminal-only coding-agent sessions are too hard to inspect. The project presents agent work as a visual tree with visible tool calls, approvals, editable context, branchable subthreads, and clients that can attach locally or remotely. Best fit: developers who want more control over Claude Code, OpenAI/Codex, Gemini, Ollama, OpenRouter, Z.ai, DeepSeek, and similar coding-agent sessions. Caveats: the author labels it alpha/beta software, the main app is AGPLv3, and teams should review network exposure before running remote sessions.
Free
Review open
A practical architecture writeup arguing that coding agents should search for reusable, maintained components before creating new code your team must own. This is a useful resource for teams building or steering coding agents. The core idea is simple: as models make code cheaper to generate, the real cost shifts to ownership, review, security, upgrades, and long-term maintenance. Use the article as a checklist before letting an agent create another internal abstraction: Ask whether the problem already has a mature library, service, API, or platform pattern. Score candidates by maintenance activity, license fit, security posture, adoption, and API stability. Generate only the smallest integration layer needed around the chosen component. Keep a decision log that explains why reuse, extension, or new code was selected. Track lifecycle cost, not only model tokens or first-pass implementation speed. The LinkLoot angle: this is not a prompt pack or a runnable tool. It is a resource for agent workflow design. It fits teams that are seeing AI-generated code volume rise faster than their ability to review, consolidate, and maintain it. Before applying it, verify the assumptions against your own codebase. Some domains need bespoke implementation for safety, compliance, performance, or product differentiation. The useful shift is not "never generate code"; it is making agents justify new code when an existing, trusted component would reduce ownership.
Free
Review open
Clawk gives Claude Code, Codex, and shell-based coding agents a disposable, network-restricted Linux VM so they can install tools and run code without direct access to your host machine. What it is Clawk is an open-source agent sandbox for local development. You start it from a repository, then run a coding agent or shell inside a disposable Linux VM with restricted outbound networking and only the mounted project files available. Why it is useful Use it when an agent needs to install packages, run servers, execute generated code, or inspect unfamiliar dependencies, but you do not want that work happening directly on your laptop. The project is pre-1.0, so treat it as a practical experiment rather than a hardened enterprise boundary. Best fit Use case Why Clawk helps Caveat --- --- --- Agent coding sessions Gives the agent root inside a throwaway Linux guest Anything mounted or allowed on the network can still be exposed Risky dependency tests Lets packages run away from the host filesystem You still need normal code review and secret hygiene Multi-agent experiments Keeps destructive commands away from the main machine Pre-1.0 project with possible breaking changes Before you try it Check the supported platforms, read the security model, and start with a non-sensitive repository. Do not mount secrets or private data unless you are comfortable with the agent and allowed network destinations seeing them.
Free
Review open
OpenAI's Codex team shared practical GPT-5.6 guidance in a Reddit AMA: use Sol Medium for most coding, reserve Sol Ultra for costly mistakes, and keep agents on bounded goals with tests. What it is A practical reference for choosing GPT-5.6 models in Codex after OpenAI's July 10 AMA. The useful signal is workflow guidance, not another launch recap: Sol Medium for most coding, Sol Ultra for migrations and security-sensitive work, Terra for faster or usage-conscious tasks, and Luna for lighter subagent work. Why bookmark it The AMA gives direct operating guidance from the Codex team instead of benchmark-only positioning. TestingCatalog distilled the model-routing notes, Codex usage comments, desktop-app friction, and persistence tips into a scan-friendly summary. The notes are useful for teams setting default reasoning levels, test requirements, and escalation rules after the GPT-5.6 rollout. Good first use Turn the AMA notes into a small routing checklist for your coding agents: task risk, repository size, required tests, acceptable latency, and when to move from Sol Medium to higher reasoning or Sol Ultra. Risk notes Treat Reddit comments and social summaries as context, not product documentation. Verify model availability, usage allowances, pricing, and context-window behavior in your own ChatGPT, Codex, or API account before changing production defaults.
Free
Review open
LongCat-2.0 is an MIT-licensed Meituan model on Hugging Face and GitHub with a 1M-token context target, coding-agent focus, and public deployment notes. Treat vendor benchmark claims as self-reported, and test it on your own repositories before trusting it in production. LongCat-2.0 is worth bookmarking if you evaluate open models for coding agents, repository-scale edits, or long-context experiments. The model card and repository describe a 1.6T-parameter MoE design, roughly 48B active parameters per token, MIT-licensed weights, a 1M-token context target, and deployment notes for SGLang/vLLM-style serving. Use it as an evaluation candidate, not an automatic production pick. The benchmark table is mostly vendor-reported, the hardware requirements are serious, and real value depends on how it handles your own codebase, tests, tool-calling format, latency, and safety controls. Practical checks before using it: Confirm the exact Hugging Face variant you want: full, FP8, INT8, or a community quantization. Run a small repository task against your current baseline model. Check license, trademark, privacy, and acceptable-use constraints for your deployment. Measure context retention and patch correctness, not only benchmark scores. Avoid assuming OpenRouter/API availability unless your provider page confirms the model at run time.
Free
Review open
A practical starter kit for evaluating Z.ai GLM-5.2 with a repository audit, a bounded refactor, and a security-review sanity check before trusting it in production. Try GLM-5.2 on a real long-context coding task GLM-5.2 is useful to test when your normal coding model loses track of repository-wide context. The practical angle is not another generic chat prompt. Use it on one bounded engineering workflow where the 1M-token context, OpenAI-compatible API access, and open-weight deployment options can be compared against your current agent stack. What to test first Start with a repository you own. Give GLM-5.2 the project structure, key docs, test commands, and one clearly scoped task. Do not begin with production write access or secrets. Use this evaluation sequence: Ask for an architecture map and risk boundaries. Run one medium refactor that should not change public APIs. Require build, lint, and test verification. Ask for a short self-review that lists files changed, assumptions, and remaining risks. Compare the result with your current primary coding model on the same task. Copyable evaluation prompt Best fit Use case Why GLM-5.2 fits Caveat --- --- --- Repository-wide audit Z.ai documents a 1M-token context and long-horizon engineering focus. Validate claims on your own codebase, not only public benchmarks. Bounded refactor The model is positioned for multi-file agentic engineering tasks. Keep API, behavior, and dependency boundaries explicit. Local/open-weight experiments Hugging Face lists the model and serving options through vLLM, SGLang, Docker Model Runner, and quantization paths. Hardware, quantization, and provider quality will change results. Security review trial Semgrep reported strong IDOR-benchmark results for GLM-5.2 under its harness. One benchmark is not proof of general security-review superiority. Access paths to compare Z.ai API: fastest path to a direct vendor test. OpenRouter: useful when you already route model calls through one API gateway. Hugging Face weights: useful for local or private serving experiments. GitHub repository: useful for release notes, model links, and serving guidance. Safety checklist Use owned or authorized repositories only. Remove secrets and customer data from prompts and logs. Keep the task bounded to one change request. Require reproducible commands, not just a confident summary. Treat benchmark wins as signals, not guarantees. Review dependency changes and generated code before merge. Source links Z.ai GLM-5.2 docs: https://docs.z.ai/guides/llm/glm-5.2 Z.ai GLM-5.2 blog: https://z.ai/blog/glm-5.2 GitHub repository: https://github.com/zai-org/GLM-5 Hugging Face model: https://huggingface.co/zai-org/GLM-5.2 OpenRouter model page: https://openrouter.ai/z-ai/glm-5.2 Semgrep benchmark context: https://semgrep.dev/blog/2026/we-have-mythos-at-home-glm-52-beats-claude-in-our-cyber-benchmarks/
Free
Review open
A premium agent workflow for creating deterministic MP4 videos from plain HTML, CSS, media, and seekable animations.
29
Review open
Blog
Related reads
AI & Automation
Alibaba releases Qwen3.8-Max for frontier coding and agent work
AI-generated · Automatically published by LinkLoot. Alibaba’s Qwen team released Qwen3.8-Max on August 3, positioning it as the family’s str…
AI & Automation
MiniMax releases M3 with 1M context and native multimodal input
MiniMax has released M3, a 428B-parameter open-weight model that combines a 1M-token context window, native text-image-video input, and agen…
AI & Automation
Anthropic releases Claude Opus 5 for everyday agent work
Anthropic has released Claude Opus 5 across Claude, Claude Code, Claude Cowork, and the Claude API, positioning it as the default Opus model…
AI & Automation
DeepSeek legacy API model names shut off on July 24
DeepSeek says the deepseek-chat and deepseek-reasoner API aliases become inaccessible on July 24, 2026 at 15:59 UTC, so agent stacks and pro…
AI & Automation
Poolside releases Laguna S 2.1 for open-weight coding agents
Poolside's Laguna S 2.1 is a 118B-parameter MoE coding-agent model with 8B active parameters, a 1M-token context window, Hugging Face weight…
AI & Automation
Meituan LongCat 2.0 reaches OpenRouter after 1.6T open model release
OpenRouter now lists Meituan LongCat 2.0, giving developers a hosted route to the 1.6T-parameter open model that LongCat documents for codin…
AI & Automation
Moonshot Kimi K3 reaches API users as older Kimi models sunset
Moonshot's Kimi K3 is live in API docs with a 2.8T-parameter, 1M-context profile, while K2.5 and moonshot-v1 close to new users and fully su…
AI & Automation
Qwen3.8-Max-Preview Reaches Alibaba Token Plan as Open Weights Wait
Alibaba's Qwen team says Qwen3.8 is coming as an open-weight 2.4T model, while the Qwen3.8-Max-Preview is already listed on Alibaba Token Pl…
Tools & Apps
GitHub adds repository-level Copilot metrics for coding agents
GitHub's Copilot usage metrics REST API now reports repository-level activity for Copilot coding agent and Copilot code review across enterp…
AI & Automation
Moonshot releases Kimi K3 with 2.8T parameters and July 27 weights
Moonshot AI has launched Kimi K3, a 2.8-trillion-parameter multimodal reasoning model available through Kimi products and API today, with fu…
AI & Automation
Inspect Grok Build Source Before Choosing an AI Coding Agent
xAI has open-sourced Grok Build, its terminal AI coding agent and TUI, giving developers a direct look at the agent loop, tools, extension s…
Tools & Apps
Cursor 3.11: Side Chats and Agent Transcript Search
Cursor 3.11 adds side chats, agent transcript search, redesigned repo pickers and cloud-agent hooks, with commands and workflow ideas for co…
AI & Automation
OpenAI says SWE-Bench Pro has a broken-task problem
OpenAI audited SWE-Bench Pro and estimates that about 30% of its public tasks are broken, changing how teams should read coding-agent benchm…
AI & Automation
Benchmark Grok 4.5 before moving coding agents off Claude or GPT
xAI has put Grok 4.5 on the API for coding, agentic tasks, and knowledge work. The useful move now is not trust the launch claims, but run y…
AI & Automation
Use Codex Mobile in ChatGPT iOS 1.2026.181 to manage coding tasks from anywhere
OpenAI's July 6 Codex changelog says ChatGPT for iOS 1.2026.181 adds stronger Codex Mobile task management, branch and change filters, trans…
AI & Automation
Run AI coding agents in Docker Sandboxes before they touch your laptop
Docker's latest agent-isolation guidance points developers to Docker Sandboxes, SBX, and Sandbox Kits for running coding agents inside dispo…
AI & Automation
Run Cursor agents from iPhone without leaving your laptop open
Cursor for iOS is now in public beta for paid plans, giving developers a mobile control surface for cloud agents, local Remote Control, revi…