{"slug": "adk", "name": "ADK", "category": "dev-libs", "category_name": "Libraries and frameworks", "description": "", "recommenders": 4, "commenters": 7, "warners": 2, "score": 2.205, "rising": 0.02, "first": "2025-11-22", "latest": "2026-09-14", "why": "I use ADK (https://google.github.io/adk-docs/runtime/) and it handles all that bookkeeping for me.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 4, "neutral": 1, "compare": 0, "warn": 2}, "spark": [0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 1, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "I use ADK (https://google.github.io/adk-docs/runtime/) and it handles all that bookkeeping for me.", "kind": "", "icon": true, "domain": "adk.dev", "github_repo": null, "official_url": "https://google.github.io/adk-docs/", "pricing": "", "aliases": ["ADK", "Google ADK"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 804449744, "item_id": 50278109, "url": "https://news.ycombinator.com/item?id=50278109", "by": "verdverm", "date": "2026-10-28", "stance": "recommend", "quote": "This is my goto SDK, it comes in multiple languages too.", "excerpt": "This is my goto SDK, it comes in multiple languages too. I'm going to be wrapping the Go ADK in CUE Nov/Dec, along with Dagger, and rebuild my personal harness from scratch which I worked on last Nov/Dec. ADK has a graph/workflow concept too: https://adk.dev/graphs/", "context": "", "use_case": "agent development", "switched_from": "", "story": "The Agent SDK"}, {"id": 795241890, "item_id": 49702618, "url": "https://news.ycombinator.com/item?id=49702618", "by": "taintech", "date": "2026-09-14", "stance": "recommend", "quote": "I use it with ADK agents, works well.", "excerpt": "Easy pattern in Cloud Run, if anyone is interested for AI agents. Use invocation based Cloud Run with Cloud Tasks. User calls service in Cloud Run, Cloud Run start Cloud Task that holds connection for 30 minutes. Cheap and easy patterns, and you avoid no cpu throttling too. I use it with ADK agents, works well.", "context": "", "use_case": "building agents", "switched_from": "", "story": "Why you can need Cloud Run Instances"}, {"id": 790128960, "item_id": 49383060, "url": "https://news.ycombinator.com/item?id=49383060", "by": "verdverm", "date": "2026-08-20", "stance": "recommend", "quote": "take a look at ADK, they have features for exactly this, mixing determ with agentic in a dag", "excerpt": "take a look at ADK, they have features for exactly this, mixing determ with agentic in a dag", "context": "", "use_case": "deterministic and agentic DAG workflows", "switched_from": "", "story": "Every Model Cheats"}, {"id": 789894112, "item_id": 49368382, "url": "https://news.ycombinator.com/item?id=49368382", "by": "verdverm", "date": "2026-08-19", "stance": "recommend", "quote": "my recommendation is https://adk.dev/", "excerpt": "Almost every harness has an SDK now, it seems part of the \"mvp\" at this point, both open and closed source https://learn.chatgpt.com/docs/codex-sdk https://code.claude.com/docs/en/agent-sdk/overview https://opencode.ai/docs/sdk/ https://pi.dev/docs/latest/sdk or if you want SDK first, my recommendation is https://adk.dev/", "context": "", "use_case": "", "switched_from": "", "story": "fx :Tiny, open, native coding agent."}, {"id": 785776640, "item_id": 49111040, "url": "https://news.ycombinator.com/item?id=49111040", "by": "verdverm", "date": "2026-07-30", "stance": "recommend", "quote": "ADK seems like a better thought through SDK family, I use Go", "excerpt": "ADK seems like a better thought through SDK family, I use Go, Py is most mature, also TS and Java. The docs still need a lot of work to cover all the languages, it's pretty hit or miss right now, but one can also have a coding agent look at the source and an example for another language. https://adk.dev/get-started/about/ | https://github.com/google/adk-go", "context": "Go user", "use_case": "building AI agents in Go", "switched_from": "", "story": "Go LLM SDK for streaming, tool-calling AI backends (plus frontend React lib)"}]}], "caveats": [{"label": "", "summary": "", "quotes": [{"id": 761858325, "item_id": 47616145, "url": "https://news.ycombinator.com/item?id=47616145", "by": "Acacian", "date": "2026-04-02", "stance": "warn", "quote": "them had it.", "excerpt": "LangChain, CrewAI, OpenAI Agents, Anthropic, LiteLLM, Pydantic AI, Google ADK \u2014 I went through 11 frameworks looking for basic runtime security: injection detection, PII masking, audit trails. None of them had it. So I built a monkey-patching layer that intercepts LLM calls and runs them through guardrails: import aegis; aegis.init() Patches whatever frameworks you have installed. ~2.6ms overhead. The nastiest find: ", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: I tested 11 AI frameworks for basic security \u2013 none passed"}, {"id": 736289968, "item_id": 46018123, "url": "https://news.ycombinator.com/item?id=46018123", "by": "cvhc", "date": "2025-11-22", "stance": "warn", "quote": "some big IDs are not passed precisely in Google ADK framework", "excerpt": "I'm frustrated that some fundamental aspects in agent development lack clear guideline. One example is input/output types of function tools. Frameworks offer some flexiblity and seemingly I can use fundamental types and simple data structures (list, dict/map). But on the other hand I know all data types are eventually stringified and this has implications. I have recently observed two issues when my agent calls a fun", "context": "", "use_case": "", "switched_from": "", "story": "Agent design is still hard"}]}], "who": [{"by": "verdverm", "context": "Go user", "url": "https://news.ycombinator.com/item?id=49111040", "date": "2026-07-30"}, {"by": "fhrow4484", "context": "Google ecosystem", "url": "https://news.ycombinator.com/item?id=48593046", "date": "2026-06-18"}, {"by": "m_manish", "context": "builds production GenAI systems in pharma, leads team of 4", "url": "https://news.ycombinator.com/item?id=47978344", "date": "2026-05-01"}], "tags": [], "use_cases": ["agent framework", "agent development", "building ai agents", "multi-language agent framework", "building agents", "deterministic and agentic dag workflows"], "compared": [{"slug": "dagger", "name": "Dagger", "co": 6, "switched_from": 0, "switched_to": 0}, {"slug": "kubernetes", "name": "Kubernetes", "co": 2, "switched_from": 0, "switched_to": 0}], "makers": [{"id": 765672882, "item_id": 47854555, "url": "https://news.ycombinator.com/item?id=47854555", "by": "verdverm", "date": "2026-04-21", "stance": "neutral", "quote": "built on ADK, CUE, and Dagger", "excerpt": "> Any other Go-based AI / LLM tools folks are happy with? I can throw my hat into the ring, built on ADK, CUE, and Dagger (all also in Go); CLI, TUI, and VSCode interfaces. It's my personal / custom stack, still need to write up docs. My favorite features are powered by Dagger, sandbox with time travel, forking, jump into shell at any turn, diff between any points. Good entrypoint folder: https://github.com/hofstadte", "context": "", "use_case": "Agent framework", "switched_from": "", "story": "Show HN: GoModel \u2013 an open-source AI gateway in Go; 44x lighter than LiteLLM"}, {"id": 762078080, "item_id": 47629880, "url": "https://news.ycombinator.com/item?id=47629880", "by": "PufPufPuf", "date": "2026-04-03", "stance": "recommend", "quote": "ADK makes it simple by using a prebuilt Lima VM for CLI deploy", "excerpt": "Actually I can, it's open source: https://github.com/kagenti/adk It's an agent development platform. It needs MicroShift, which is notoriously not easy to run locally -- ADK makes it simple by using a prebuilt Lima VM for CLI deploy, but that doesn't work well in development.", "context": "", "use_case": "Agent development platform", "switched_from": "", "story": "Show HN: Locki \u2013 AI sandboxing for real-world projects"}, {"id": 754415296, "item_id": 47150956, "url": "https://news.ycombinator.com/item?id=47150956", "by": "mbleterman", "date": "2026-02-25", "stance": "neutral", "quote": "this uses Google's ADK to run specialized agents", "excerpt": "Author here. Happy to answer questions about the implementation. Quick context: this uses Google's ADK to run specialized agents (Backend, Frontend, QA, Security, PM) coordinated by an Orchestrator. They execute sprints in parallel with a defect detection loop. Environment setup overhead was brutal (40%+ of tokens). But the QA loop catching regressions was worth it. GitHub: https://github.com/michaelbleterman/ai-scru", "context": "", "use_case": "Multi-agent framework", "switched_from": "", "story": "AI-Scrum: Applying Scrum methodology to AI agent teams"}], "threads": [{"url": "https://news.ycombinator.com/item?id=46013935", "title": "Agent design is still hard", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=47628034", "title": "Show HN: Locki \u2013 AI sandboxing for real-world projects", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=46203824", "title": "SSE sucks for transporting LLM tokens", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=50273731", "title": "The Agent SDK", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49839567", "title": "Evolving programming languages in the AI era", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49701509", "title": "Why you can need Cloud Run Instances", "mentions": 1}], "sentiment_by_month": {"2025-11": {"warn": 1}, "2025-12": {}, "2026-01": {}, "2026-02": {"recommend": 1}, "2026-03": {}, "2026-04": {"warn": 1, "recommend": 1}, "2026-05": {"neutral": 1}, "2026-06": {"recommend": 1}, "2026-07": {}, "2026-08": {}, "2026-09": {"recommend": 1}, "2026-10": {}}, "quotes": [{"id": 804449744, "item_id": 50278109, "url": "https://news.ycombinator.com/item?id=50278109", "by": "verdverm", "date": "2026-10-28", "stance": "recommend", "quote": "This is my goto SDK, it comes in multiple languages too.", "excerpt": "This is my goto SDK, it comes in multiple languages too. I'm going to be wrapping the Go ADK in CUE Nov/Dec, along with Dagger, and rebuild my personal harness from scratch which I worked on last Nov/Dec. ADK has a graph/workflow concept too: https://adk.dev/graphs/", "context": "", "use_case": "agent development", "switched_from": "", "story": "The Agent SDK"}, {"id": 797792976, "item_id": 49862061, "url": "https://news.ycombinator.com/item?id=49862061", "by": "verdverm", "date": "2026-09-27", "stance": "neutral", "quote": "here are two I use", "excerpt": "maybe more like an ecosystem around a framework implemented in multiple language SDKs, here are two I use https://adk.dev/ https://docs.dagger.io/reference/sdks both can invoke modules written in other languages from your language of choice Kubernetes is likely an interesting ecosystem to consider under this lens too", "context": "", "use_case": "multi-language agent framework", "switched_from": "", "story": "Evolving programming languages in the AI era"}, {"id": 795241890, "item_id": 49702618, "url": "https://news.ycombinator.com/item?id=49702618", "by": "taintech", "date": "2026-09-14", "stance": "recommend", "quote": "I use it with ADK agents, works well.", "excerpt": "Easy pattern in Cloud Run, if anyone is interested for AI agents. Use invocation based Cloud Run with Cloud Tasks. User calls service in Cloud Run, Cloud Run start Cloud Task that holds connection for 30 minutes. Cheap and easy patterns, and you avoid no cpu throttling too. I use it with ADK agents, works well.", "context": "", "use_case": "building agents", "switched_from": "", "story": "Why you can need Cloud Run Instances"}, {"id": 790128960, "item_id": 49383060, "url": "https://news.ycombinator.com/item?id=49383060", "by": "verdverm", "date": "2026-08-20", "stance": "recommend", "quote": "take a look at ADK, they have features for exactly this, mixing determ with agentic in a dag", "excerpt": "take a look at ADK, they have features for exactly this, mixing determ with agentic in a dag", "context": "", "use_case": "deterministic and agentic DAG workflows", "switched_from": "", "story": "Every Model Cheats"}, {"id": 789894112, "item_id": 49368382, "url": "https://news.ycombinator.com/item?id=49368382", "by": "verdverm", "date": "2026-08-19", "stance": "recommend", "quote": "my recommendation is https://adk.dev/", "excerpt": "Almost every harness has an SDK now, it seems part of the \"mvp\" at this point, both open and closed source https://learn.chatgpt.com/docs/codex-sdk https://code.claude.com/docs/en/agent-sdk/overview https://opencode.ai/docs/sdk/ https://pi.dev/docs/latest/sdk or if you want SDK first, my recommendation is https://adk.dev/", "context": "", "use_case": "", "switched_from": "", "story": "fx :Tiny, open, native coding agent."}, {"id": 785776640, "item_id": 49111040, "url": "https://news.ycombinator.com/item?id=49111040", "by": "verdverm", "date": "2026-07-30", "stance": "recommend", "quote": "ADK seems like a better thought through SDK family, I use Go", "excerpt": "ADK seems like a better thought through SDK family, I use Go, Py is most mature, also TS and Java. The docs still need a lot of work to cover all the languages, it's pretty hit or miss right now, but one can also have a coding agent look at the source and an example for another language. https://adk.dev/get-started/about/ | https://github.com/google/adk-go", "context": "Go user", "use_case": "building AI agents in Go", "switched_from": "", "story": "Go LLM SDK for streaming, tool-calling AI backends (plus frontend React lib)"}, {"id": 779800065, "item_id": 48737504, "url": "https://news.ycombinator.com/item?id=48737504", "by": "verdverm", "date": "2026-06-30", "stance": "neutral", "quote": "I used ADK, Dagger, and a VS Code extension for mine.", "excerpt": "Lots of ways, it's a good exercise that you will learn a lot doing. Might make you cynical w.r.t. big ai harnesses I used ADK, Dagger, and a VS Code extension for mine. Currently using opencode though.", "context": "", "use_case": "building personal harness", "switched_from": "", "story": "Claude Code Is Steganographically Marking Requests"}, {"id": 777488737, "item_id": 48593046, "url": "https://news.ycombinator.com/item?id=48593046", "by": "fhrow4484", "date": "2026-06-18", "stance": "recommend", "quote": "can expose a custom Google ADK \"agent\" via one-liner as a2a", "excerpt": "If you're big in the google ecosystem it makes sense: - gemini-cli can have remote a2a agents connected to it. (See https://github.com/tanaikech/gemini-cli-gas-a2a-subagents and the medium article linked in that repo description.) Haven't tried it but I tried the local use case below. - can expose a custom Google ADK \"agent\" via one-liner as a2a, which you can connect to the gemini-cli the same way.", "context": "Google ecosystem", "use_case": "exposing agents via A2A", "switched_from": "", "story": "Ask HN: Is anyone using the A2A protocol yet?"}, {"id": 767653506, "item_id": 47978344, "url": "https://news.ycombinator.com/item?id=47978344", "by": "m_manish", "date": "2026-05-01", "stance": "neutral", "quote": "GenAI: LangGraph, AutoGen, Google ADK, RAG systems", "excerpt": "Location: India Remote: Yes (preferred) Willing to relocate: No Technologies: GenAI: LangGraph, AutoGen, Google ADK, RAG systems Backend/Data: Python, FastAPI, PostgreSQL, DuckDB Cloud: Azure, Databricks, Docker, Kubernetes R\u00e9sum\u00e9: https://manish-m.com/resume.pdf Email: workdesk <dot> manish [at] gmail <dot> com I build and ship production GenAI systems in pharma (R&D + TechOps), leading a team of 4. My work spans da", "context": "builds production GenAI systems in pharma, leads team of 4", "use_case": "", "switched_from": "", "story": "Ask HN: Who wants to be hired? (May 2026)"}, {"id": 763001168, "item_id": 47687573, "url": "https://news.ycombinator.com/item?id=47687573", "by": "harlequinetcie", "date": "2026-04-08", "stance": "recommend", "quote": "ADK was (and is) exceptional", "excerpt": "Oh Google, I love you guys, but it seems they alway launch these half baked things without the support they deserve behind it. ADK was (and is) exceptional, but nobody is actually making noise and pushing for it as they should. It feels like Microsoft .net back in the day. Let's see how it goes. I'm rooting for y'all", "context": "", "use_case": "agent development", "switched_from": "", "story": "Google Open Sources Experimental Agent Orchestration Testbed Scion"}, {"id": 762537888, "item_id": 47658618, "url": "https://news.ycombinator.com/item?id=47658618", "by": "PufPufPuf", "date": "2026-04-06", "stance": "neutral", "quote": "ADK is quite \"early\" in development", "excerpt": "To complicate it further, ADK is quite \"early\" in development, but it's a fusion-fork of two more mature projects: Red Hat's Kagenti, and Linux Foundation's (orig. IBM) Agent Stack (https://github.com/i-am-bee/agentstack/). ADK is basically Agent Stack but with the custom backend replaced by Kagenti. Regarding MCP: the current idea is dependency injection using A2A extensions: agent says what it wants (LLM endpoint, ", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Locki \u2013 AI sandboxing for real-world projects"}, {"id": 761858325, "item_id": 47616145, "url": "https://news.ycombinator.com/item?id=47616145", "by": "Acacian", "date": "2026-04-02", "stance": "warn", "quote": "them had it.", "excerpt": "LangChain, CrewAI, OpenAI Agents, Anthropic, LiteLLM, Pydantic AI, Google ADK \u2014 I went through 11 frameworks looking for basic runtime security: injection detection, PII masking, audit trails. None of them had it. So I built a monkey-patching layer that intercepts LLM calls and runs them through guardrails: import aegis; aegis.init() Patches whatever frameworks you have installed. ~2.6ms overhead. The nastiest find: ", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: I tested 11 AI frameworks for basic security \u2013 none passed"}, {"id": 761238160, "item_id": 47577385, "url": "https://news.ycombinator.com/item?id=47577385", "by": "verdverm", "date": "2026-03-30", "stance": "recommend", "quote": "Big fan of ADK, I use the Go version.", "excerpt": "Big fan of ADK, I use the Go version. There are also Typescript and Python flavors", "context": "", "use_case": "", "switched_from": "", "story": "ADK for Java 1.0.0: Building the Future of AI Agents in Java"}, {"id": 759855152, "item_id": 47490947, "url": "https://news.ycombinator.com/item?id=47490947", "by": "verdverm", "date": "2026-03-23", "stance": "recommend", "quote": "I use ADK, haven't really looked at Dspy", "excerpt": "Have you looked at ADK? How does it compare? Does it even fit in the same place as Dspy? https://google.github.io/adk-docs/ Disclaimer, I use ADK, haven't really looked at Dspy (though I have prior heard of it). ADK certainly addresses all of the points you have in the post.", "context": "", "use_case": "agent development", "switched_from": "", "story": "If Dspy is so great, why isn't anyone using it?"}, {"id": 757881600, "item_id": 47367600, "url": "https://news.ycombinator.com/item?id=47367600", "by": "verdverm", "date": "2026-03-13", "stance": "recommend", "quote": "The framework I use (ADK) already handles this, very low hanging fruit", "excerpt": "I don't want some other tooling messing with my context. It's too important to leave to something that needs to optimize across many users, there by not being the best for my specifics. The framework I use (ADK) already handles this, very low hanging fruit that should be a part of any framework, not something external. In ADK, this is a boolean you can turn on per tool or subagent, you can even decide turn by turn or", "context": "", "use_case": "agent context management", "switched_from": "", "story": ""}, {"id": 757880256, "item_id": 47367516, "url": "https://news.ycombinator.com/item?id=47367516", "by": "verdverm", "date": "2026-03-13", "stance": "recommend", "quote": "I use ADK, a well thought through framework", "excerpt": "I use ADK, a well thought through framework by a company that leads in both AI and engineering. They are the only company in existence with the experience in both to build a solid, production worthy agent framework. People too focused on AI and what's popular are missing the forest for a tree. Most of the things I see being talked about as problems & solutions have already been taken care of with mature tooling choic", "context": "", "use_case": "agent framework", "switched_from": "", "story": "How are people debugging multi-agent AI workflows in production?"}, {"id": 754313505, "item_id": 47144594, "url": "https://news.ycombinator.com/item?id=47144594", "by": "verdverm", "date": "2026-02-24", "stance": "recommend", "quote": "versus something like ADK, developed by many people and backed by Google", "excerpt": "Can you tell my why you think I should choose a project started yesterday[1] by one person (driving an ai), versus something like ADK, developed by many people and backed by Google? [1] https://github.com/MaxGfeller/open-harness/commit/82019d703264c9f08be66b4363d070f029a4bf66", "context": "", "use_case": "", "switched_from": "", "story": "OpenHarness: A code-first, composable SDK to build powerful AI agents"}, {"id": 754248736, "item_id": 47140546, "url": "https://news.ycombinator.com/item?id=47140546", "by": "verdverm", "date": "2026-02-24", "stance": "neutral", "quote": "I use ADK which has many points 3rd parties plug in.", "excerpt": "I use ADK which has many points 3rd parties plug in. I'm also involved in the development (from the outside). I will look more into Pilo and how this could work. Would save me a bunch of effort! I'll open an issue for tracking --- said issue: https://github.com/mozilla/pilo/issues/318", "context": "involved in development from the outside", "use_case": "agent framework", "switched_from": "", "story": "Show HN: Pilo \u2013 open-source agentic web automation engine by Mozilla"}, {"id": 753681120, "item_id": 47105070, "url": "https://news.ycombinator.com/item?id=47105070", "by": "verdverm", "date": "2026-02-21", "stance": "recommend", "quote": "I like ADK, it's lower level and more general", "excerpt": "I like ADK, it's lower level and more general, so there is a bit you have to do to get a \"claw\" like experience (not that much) and you get (1) a common framework you can use for other things (2) a lot more places to plug in (3) four SDKs to choose from (ts, go, py, java... so far) It's a lot more work to build a Copilot alternative (ide integration, cli). I've done a lot of that with adk-go, https://github.com/hofst", "context": "", "use_case": "", "switched_from": "", "story": "Andrej Karpathy talks about \"Claws\""}, {"id": 752753984, "item_id": 47047124, "url": "https://news.ycombinator.com/item?id=47047124", "by": "stsffap", "date": "2026-02-17", "stance": "recommend", "quote": "ADK is great framework for building agents as it is runtime agnostic", "excerpt": "ADK is great framework for building agents as it is runtime agnostic and you can choose which properties are important to you. You can run your ADK agents also on Restate (https://google.github.io/adk-docs/integrations/restate) if you want to turn your agents into durable agents that can reliably communicate with other agents.", "context": "", "use_case": "building AI agents", "switched_from": "", "story": "A remote control for your agents"}, {"id": 752720816, "item_id": 47045051, "url": "https://news.ycombinator.com/item?id=47045051", "by": "verdverm", "date": "2026-02-17", "stance": "neutral", "quote": "ADK framework I use handles parallel agents/tools.", "excerpt": "Is this more than keeping history before compaction, making it all available via tools, and some backend bookkeeping for parallel tool calls? I'm not exactly sure what is interesting here Much of this feels like a technical report of what they did, and makes me feel like we've reached the ICO whitepaper phase. I have very similar features in my custom coding agent, they seem pretty common sense to have. Are you reall", "context": "", "use_case": "custom coding agent", "switched_from": "", "story": "LCM: Lossless Context Management"}, {"id": 752482512, "item_id": 47030157, "url": "https://news.ycombinator.com/item?id=47030157", "by": "verdverm", "date": "2026-02-15", "stance": "recommend", "quote": "For Ai, I'm using ADK.", "excerpt": "There are a number of other aspects that make kubernetes what it is. In particular the API surface is well defined at a very good abstraction level. For Ai, I'm using ADK. It's my \"kubernetes for agents\" from this perspective of abstractions and common task management handled by the system. https://google.github.io/adk-docs/get-started/about/", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Kubernetes for AI Agents"}, {"id": 751993360, "item_id": 46999585, "url": "https://news.ycombinator.com/item?id=46999585", "by": "verdverm", "date": "2026-02-12", "stance": "recommend", "quote": "I'd look to Google's ADK for what enterprise open source / production features look like.", "excerpt": "I'd look to Google's ADK for what enterprise open source / production features look like. Available in { python, java, golang, typescript } at varying levels of maturity. Work appears rapid (I maintain a fork for adk-go, which includes sandboxing, time travel, and a few bug fixes / enhancements). Great interfaces that are easy to extend with code. https://google.github.io/adk-docs/get-started/about", "context": "", "use_case": "Agent framework", "switched_from": "", "story": "Ask HN: What makes an AI agent framework production-ready vs. a toy?"}, {"id": 751855696, "item_id": 46990981, "url": "https://news.ycombinator.com/item?id=46990981", "by": "verdverm", "date": "2026-02-12", "stance": "recommend", "quote": "I use ADK and don't have this problem", "excerpt": "I use ADK and don't have this problem, I can switch sessions in my coding agent and they keep running in the background. If a new event comes in, and I have that session open, then the UI updates This isn't really a problem, just bad design on the apps part. tl;dr - don't intertwine your agent engine with UI rendering, easy enough?", "context": "", "use_case": "building coding agents", "switched_from": "", "story": "A chatbot's worst enemy is page refresh"}, {"id": 749472401, "item_id": 46842025, "url": "https://news.ycombinator.com/item?id=46842025", "by": "verdverm", "date": "2026-01-31", "stance": "neutral", "quote": "The ADK framework I use does most of the rest", "excerpt": "I solved this with Dagger and OCI layers, so I get all these same features while also doing so in an isolated environment in a ubiquitous format. The ADK framework I use does most of the rest Not a prod time thing, unless you consider a coding agent using this as prod, because it's the real thing at development time", "context": "", "use_case": "agent framework", "switched_from": "", "story": "Show HN: I built a small OSS kernel for replaying and diffing AI decisions"}, {"id": 742339536, "item_id": 46396221, "url": "https://news.ycombinator.com/item?id=46396221", "by": "verdverm", "date": "2025-12-26", "stance": "recommend", "quote": "That's why I use ADK anyway", "excerpt": "What you are describing is context construction. When working with agents and LLMs, the only way you get anything beyond their training is through the system prompt and message history. There is nothing else. You can call them whatever fancy notions and anthropomorphic concepts you want, but in the end, it is just context engineering, regardless of how and when you create, store, retrieve, and inject artifacts. A goo", "context": "", "use_case": "Agent framework", "switched_from": "", "story": "Why do so many \"agentic AI\" systems collapse without persistent state?"}, {"id": 740465808, "item_id": 46279113, "url": "https://news.ycombinator.com/item?id=46279113", "by": "verdverm", "date": "2025-12-15", "stance": "recommend", "quote": "ADK is open source as I like it", "excerpt": "The three options you offered are all VC funded \"open source\" This is where the real AI bubble is. VC funded startups who's main plan is likely an acquisition. I'm not interested in those kinds of \"open source\" anymore, they want to lock you in to their product. Also, https://ssotax.org/ ADK is open source as I like it", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: What Are You Working On? (December 2025)"}, {"id": 740140864, "item_id": 46258804, "url": "https://news.ycombinator.com/item?id=46258804", "by": "verdverm", "date": "2025-12-13", "stance": "recommend", "quote": "the ADK framework I use handles all that bookkeeping for me", "excerpt": "yup, the ADK framework I use handles all that bookkeeping for me, has a few options built in, and pretty easy to add new implementations. I'm extending it to attach filesys+exe changes during a session, persisted as container layers via Dagger", "context": "", "use_case": "Agent session bookkeeping", "switched_from": "", "story": "SSE sucks for transporting LLM tokens"}, {"id": 740121312, "item_id": 46257582, "url": "https://news.ycombinator.com/item?id=46257582", "by": "verdverm", "date": "2025-12-13", "stance": "recommend", "quote": "I use ADK (https://google.github.io/adk-docs/runtime/) and it handles all that bookkeeping for me.", "excerpt": "Exactly, I use ADK (https://google.github.io/adk-docs/runtime/) and it handles all that bookkeeping for me. If my client disconnects, the engine continues to run until the turn is up, recording all events. We need more frameworks like ADK that handle the bookkeeping and give us great abstractions for building agentic systems.", "context": "", "use_case": "building agentic systems with resumable streaming", "switched_from": "", "story": "SSE sucks for transporting LLM tokens"}, {"id": 736289968, "item_id": 46018123, "url": "https://news.ycombinator.com/item?id=46018123", "by": "cvhc", "date": "2025-11-22", "stance": "warn", "quote": "some big IDs are not passed precisely in Google ADK framework", "excerpt": "I'm frustrated that some fundamental aspects in agent development lack clear guideline. One example is input/output types of function tools. Frameworks offer some flexiblity and seemingly I can use fundamental types and simple data structures (list, dict/map). But on the other hand I know all data types are eventually stringified and this has implications. I have recently observed two issues when my agent calls a fun", "context": "", "use_case": "", "switched_from": "", "story": "Agent design is still hard"}, {"id": 736286512, "item_id": 46017907, "url": "https://news.ycombinator.com/item?id=46017907", "by": "verdverm", "date": "2025-11-22", "stance": "recommend", "quote": "Generally very optimistic that I found a solid foundation and framework", "excerpt": "w.r.t. Go, it's probably not that big a lift. I was looking at that yesterday, made a small change to lift the Gorm stuff a bit so the DB conn can be shared between the services I thought the same thing about the artifact service, which could have a nice local FS option. I'm pretty new to ADK, so we'll see how long the honeymoon phase lasts. Generally very optimistic that I found a solid foundation and framework edit", "context": "Pretty new to ADK", "use_case": "Agent framework in Go", "switched_from": "", "story": "Agent design is still hard"}, {"id": 736279984, "item_id": 46017499, "url": "https://news.ycombinator.com/item?id=46017499", "by": "verdverm", "date": "2025-11-22", "stance": "neutral", "quote": "ADK has a few pages and some API for evaluating agentic systems", "excerpt": "ADK has a few pages and some API for evaluating agentic systems https://google.github.io/adk-docs/evaluate/ tl;dr - challenging because different runs produce different output, also how do you pass/fail (another LLM/agent is what people do)", "context": "", "use_case": "agent evaluation", "switched_from": "", "story": "Agent design is still hard"}, {"id": 736031696, "item_id": 46001981, "url": "https://news.ycombinator.com/item?id=46001981", "by": "verdverm", "date": "2025-11-20", "stance": "recommend", "quote": "am quite impressed after a week of test driving", "excerpt": "I've been using https://google.github.io/adk-docs/ and am quite impressed after a week of test driving", "context": "", "use_case": "building AI agents", "switched_from": "", "story": "Ask HN: Best solution to build AI agents?"}]}