{"slug": "cline", "name": "Cline", "category": "ai-coding", "category_name": "AI coding tools", "description": "Open-source AI coding agent extension for VS Code that works with many model providers.", "recommenders": 8, "commenters": 17, "warners": 2, "score": 6.81, "rising": 0.04, "first": "2025-11-02", "latest": "2026-10-26", "why": "Liked as a VS Code coding agent that works with many models and providers, and several commenters say it outperforms GitHub Copilot.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 8, "neutral": 7, "compare": 1, "warn": 2}, "spark": [2, 0, 2, 0, 1, 0, 0, 0, 1, 1, 0, 1], "listed": true, "loved": ["Model choice", "Output quality"], "caveat": "Ignores plan mode", "short_quote": "I\u2019ll express my personal preference for VS Code with Cline.", "icon": true, "domain": null, "github_repo": null, "official_url": "https://cline.bot", "pricing": "open source", "aliases": ["Cline"], "verdict": [{"text": "Liked as a VS Code coding agent that works with many models and providers,", "cites": [{"id": 744088992, "item_id": 46505562, "url": "https://news.ycombinator.com/item?id=46505562", "by": "masto", "date": "2026-01-05", "stance": "recommend", "quote": "I\u2019ll express my personal preference for VS Code with Cline.", "excerpt": "I\u2019ll express my personal preference for VS Code with Cline. I don\u2019t know exactly why, but its workflow feels right even though they\u2019re all almost the same interface. I do like the huge choice of models and payment options. For something I use burstily, it makes sense for me to pay as I go.", "context": "burst usage", "use_case": "AI coding, pay as you go", "switched_from": "", "story": "I came back from Cursor to VS Code"}, {"id": 735518770, "item_id": 45969923, "url": "https://news.ycombinator.com/item?id=45969923", "by": "syspec", "date": "2025-11-18", "stance": "neutral", "quote": "I use Cline, it's just a dropdown change", "excerpt": "I have \"unlimited\" access to both Gemini 2.5 Pro and Claude 4.5 Sonnet through work. From my experience, both are capable and can solve nearly all the same complex programming requests, but time and time again Gemini spits out reams and reams of code so over engineered, that totally works, but I would never want to have to interact with. When looking at the code, you can't tell why it looks \"gross\", but then you ask ", "context": "has work access to both models", "use_case": "AI coding", "switched_from": "", "story": "Gemini 3"}, {"id": 804035377, "item_id": 50252211, "url": "https://news.ycombinator.com/item?id=50252211", "by": "anigbrowl", "date": "2026-10-26", "stance": "recommend", "quote": "I had moderately good results with using that as a coding agent", "excerpt": "I'm inclined to agree. Jetbrains' own agent host 'Junie' was such a paim to set up and get working with other providers that I switched to Cline. I had moderately good results with using that as a coding agent, but eventually I switched to Deepseek Harness for generative tasks and am now using my IDEs more for review.", "context": "", "use_case": "AI coding agent", "switched_from": "Junie", "story": "JetBrains Thinkrail"}]}, {"text": " and several commenters say it outperforms GitHub Copilot.", "cites": [{"id": 745241493, "item_id": 46577593, "url": "https://news.ycombinator.com/item?id=46577593", "by": "kasey_junk", "date": "2026-01-11", "stance": "recommend", "quote": "Several of the open source agents will out perform copilot regularly (try aider, open code or cline if you can).", "excerpt": "It is not the same. The agents (Claude code, codex) are the actual processes you\u2019d interact with instead of copilot. The choices those processes make about tool calls, using subagents, etc make a huge difference in the quality of result you\u2019ll get. Copilot is just an extremely bad agent compared to the sota agents. Several of the open source agents will out perform copilot regularly (try aider, open code or cline if ", "context": "", "use_case": "AI coding agent", "switched_from": "", "story": "Ask HN: Senior software engineers, how do you use Claude Code?"}, {"id": 787562305, "item_id": 49222644, "url": "https://news.ycombinator.com/item?id=49222644", "by": "vijucat", "date": "2026-08-08", "stance": "recommend", "quote": "you have to use Cline (or other harness) if using vscode.", "excerpt": "Caching makes a huge difference to cost. On Fireworks AI, for example, if it hits the cache, you pay only 20%. And uncached is just $0.14/M tokens for DSV4-0731! I get entire re-architecture projects (with new tests and documentation) done for mere dollars. DSV4-0731 is a daily driver for me. But note that you have to use Cline (or other harness) if using vscode. I was shocked at how poor the recent versions of GitHu", "context": "", "use_case": "coding agent harness in VS Code", "switched_from": "", "story": "DeepSeek V4 Flash 0731"}]}, {"text": " Caveats include the agent making changes while plan mode was on,", "cites": [{"id": 802427857, "item_id": 50151741, "url": "https://news.ycombinator.com/item?id=50151741", "by": "SftwareRenderer", "date": "2026-10-19", "stance": "warn", "quote": "I originally made this out of frustration when I used to use Cline (the agent would make changes even when their plan mode was enabled)", "excerpt": "It depends on the operating mode you choose and your prompt/AGENTS.md file. The \"Act\" operating mode has strict guardrails, and a card only moves into review if the agent explicitly asks for your review (via tool call) or has a question. It's less prone to unnecessary interruptions. For example agents often decide to pause during a multi-phase plan even though no human input is needed. If an agent decides to pause in", "context": "", "use_case": "coding agent", "switched_from": "", "story": "Show HN: Cosplay as a tech lead, using gVisor to sandbox agents"}]}, {"text": " repeated mistakes on a poorly documented framework migration,", "cites": [{"id": 777933040, "item_id": 48620815, "url": "https://news.ycombinator.com/item?id=48620815", "by": "foobarchu", "date": "2026-06-21", "stance": "warn", "quote": "I used Claude-backed Cline in March to make that migration, and it got SO much wrong", "excerpt": "> Migrating Spring Boot apps from 3.x t 4.x is now easy given all the tooling available. Is this true now? I used Claude-backed Cline in March to make that migration, and it got SO much wrong, mainly due to the lack of solid documentation and example code for it, I imagine. Probably the project where I've had to fix the most AI mistakes so far (it was incapable of fixing the mistakes itself, it just kept getting it w", "context": "", "use_case": "Spring Boot migration", "switched_from": "", "story": "CTOs Agree: Cognitive Debt Is the New Technical Debt"}]}, {"text": " and costs that climb as context grows or with pricier models.", "cites": [{"id": 763837188, "item_id": 47739824, "url": "https://news.ycombinator.com/item?id=47739824", "by": "rzkyif", "date": "2026-04-12", "stance": "neutral", "quote": "I also tried VS Code + Cline + OpenRouter + MiniMax M2.7.", "excerpt": "My personal experience is way different: I struggle to burn through more than 50% of the 5 hour limit For context, with Google AI Pro, I can burn through the Antigravity weekly limit in 1-2 hours if I force it to use Gemini 3.1 Pro. Meanwhile Gemini 3 Flash is basically unlimited but frequently produces buggy code or fail to implement things how I personally would (felt like it doesn't \"think\" like a software dev) I ", "context": "", "use_case": "AI coding in VS Code", "switched_from": "", "story": "Pro Max 5x Quota Exhausted in 1.5 Hours Despite Moderate Usage"}, {"id": 732653251, "item_id": 45790828, "url": "https://news.ycombinator.com/item?id=45790828", "by": "r_singh", "date": "2025-11-02", "stance": "recommend", "quote": "Anthropic models via OpenRouter + Cline \u2013 when the task is error-prone, tedious, or needs high fidelity", "excerpt": "Here\u2019s how I decide what to use: ChatGPT \u2013 journaling, talking, planning. Codex \u2013 framework and middleware-layer coding. Claude Code \u2013 logic and application-level coding. Anthropic models via OpenRouter + Cline \u2013 when the task is error-prone, tedious, or needs high fidelity; lower error rate in my experience, though pricier. Cursor Agents \u2013 multi-file integration, boilerplate, and forking tasks. Each fills a differen", "context": "", "use_case": "high-fidelity coding tasks", "switched_from": "", "story": "Ask HN: Seriously, which AI coding thing is best?"}]}], "praise": [{"label": "Model choice", "summary": "Commenters value the wide choice of models and providers, with switching described as just a dropdown change.", "quotes": [{"id": 744088992, "item_id": 46505562, "url": "https://news.ycombinator.com/item?id=46505562", "by": "masto", "date": "2026-01-05", "stance": "recommend", "quote": "I\u2019ll express my personal preference for VS Code with Cline.", "excerpt": "I\u2019ll express my personal preference for VS Code with Cline. I don\u2019t know exactly why, but its workflow feels right even though they\u2019re all almost the same interface. I do like the huge choice of models and payment options. For something I use burstily, it makes sense for me to pay as I go.", "context": "burst usage", "use_case": "AI coding, pay as you go", "switched_from": "", "story": "I came back from Cursor to VS Code"}, {"id": 735518770, "item_id": 45969923, "url": "https://news.ycombinator.com/item?id=45969923", "by": "syspec", "date": "2025-11-18", "stance": "neutral", "quote": "I use Cline, it's just a dropdown change", "excerpt": "I have \"unlimited\" access to both Gemini 2.5 Pro and Claude 4.5 Sonnet through work. From my experience, both are capable and can solve nearly all the same complex programming requests, but time and time again Gemini spits out reams and reams of code so over engineered, that totally works, but I would never want to have to interact with. When looking at the code, you can't tell why it looks \"gross\", but then you ask ", "context": "has work access to both models", "use_case": "AI coding", "switched_from": "", "story": "Gemini 3"}, {"id": 804035377, "item_id": 50252211, "url": "https://news.ycombinator.com/item?id=50252211", "by": "anigbrowl", "date": "2026-10-26", "stance": "recommend", "quote": "I had moderately good results with using that as a coding agent", "excerpt": "I'm inclined to agree. Jetbrains' own agent host 'Junie' was such a paim to set up and get working with other providers that I switched to Cline. I had moderately good results with using that as a coding agent, but eventually I switched to Deepseek Harness for generative tasks and am now using my IDEs more for review.", "context": "", "use_case": "AI coding agent", "switched_from": "Junie", "story": "JetBrains Thinkrail"}]}, {"label": "Output quality", "summary": "Users report it works well out of the box, one-shots apps and beats Copilot.", "quotes": [{"id": 733714080, "item_id": 45857130, "url": "https://news.ycombinator.com/item?id=45857130", "by": "sheepscreek", "date": "2025-11-08", "stance": "recommend", "quote": "", "excerpt": "I used their $50 plan and with the previously offered Qwen3 coder 480B. While fast - none of the \u201csupported\u201d tools I tried were able to use it in a way that didn\u2019t hit the per minute request limit in a few seconds. It was incredibly frustrating. For the record, I tried OpenCoder, VSCode, Quen Coder CLI, octofriend and a few others I don\u2019t remember. Fast forward to now, when GLM 4.6 has replaced Qwen3 coder in their s", "context": "", "use_case": "AI coding with Cerebras API", "switched_from": "", "story": "Cerebras Code now supports GLM 4.6 at 1000 tokens/sec"}, {"id": 756092496, "item_id": 47255781, "url": "https://news.ycombinator.com/item?id=47255781", "by": "diebillionaires", "date": "2026-03-04", "stance": "recommend", "quote": "i can get an app one shotted pretty much every time i use cline too.", "excerpt": "i can get an app one shotted pretty much every time i use cline too. why would i ever use a proprietary solution that\u2019s opinionated and likely eventually costs (idk i didn't look)? add this to the bubble theory", "context": "", "use_case": "generating apps", "switched_from": "", "story": "Glaze by Raycast"}, {"id": 745241493, "item_id": 46577593, "url": "https://news.ycombinator.com/item?id=46577593", "by": "kasey_junk", "date": "2026-01-11", "stance": "recommend", "quote": "Several of the open source agents will out perform copilot regularly (try aider, open code or cline if you can).", "excerpt": "It is not the same. The agents (Claude code, codex) are the actual processes you\u2019d interact with instead of copilot. The choices those processes make about tool calls, using subagents, etc make a huge difference in the quality of result you\u2019ll get. Copilot is just an extremely bad agent compared to the sota agents. Several of the open source agents will out perform copilot regularly (try aider, open code or cline if ", "context": "", "use_case": "AI coding agent", "switched_from": "", "story": "Ask HN: Senior software engineers, how do you use Claude Code?"}]}, {"label": "Cost control", "summary": "Commenters credit it with using prompt caching well, and one likes paying as they go.", "quotes": [{"id": 787562305, "item_id": 49222644, "url": "https://news.ycombinator.com/item?id=49222644", "by": "vijucat", "date": "2026-08-08", "stance": "recommend", "quote": "you have to use Cline (or other harness) if using vscode.", "excerpt": "Caching makes a huge difference to cost. On Fireworks AI, for example, if it hits the cache, you pay only 20%. And uncached is just $0.14/M tokens for DSV4-0731! I get entire re-architecture projects (with new tests and documentation) done for mere dollars. DSV4-0731 is a daily driver for me. But note that you have to use Cline (or other harness) if using vscode. I was shocked at how poor the recent versions of GitHu", "context": "", "use_case": "coding agent harness in VS Code", "switched_from": "", "story": "DeepSeek V4 Flash 0731"}, {"id": 744088992, "item_id": 46505562, "url": "https://news.ycombinator.com/item?id=46505562", "by": "masto", "date": "2026-01-05", "stance": "recommend", "quote": "I\u2019ll express my personal preference for VS Code with Cline.", "excerpt": "I\u2019ll express my personal preference for VS Code with Cline. I don\u2019t know exactly why, but its workflow feels right even though they\u2019re all almost the same interface. I do like the huge choice of models and payment options. For something I use burstily, it makes sense for me to pay as I go.", "context": "burst usage", "use_case": "AI coding, pay as you go", "switched_from": "", "story": "I came back from Cursor to VS Code"}]}, {"label": "Lower error rate", "summary": "One user picks it with Anthropic models for error-prone, high-fidelity tasks.", "quotes": [{"id": 732653251, "item_id": 45790828, "url": "https://news.ycombinator.com/item?id=45790828", "by": "r_singh", "date": "2025-11-02", "stance": "recommend", "quote": "Anthropic models via OpenRouter + Cline \u2013 when the task is error-prone, tedious, or needs high fidelity", "excerpt": "Here\u2019s how I decide what to use: ChatGPT \u2013 journaling, talking, planning. Codex \u2013 framework and middleware-layer coding. Claude Code \u2013 logic and application-level coding. Anthropic models via OpenRouter + Cline \u2013 when the task is error-prone, tedious, or needs high fidelity; lower error rate in my experience, though pricier. Cursor Agents \u2013 multi-file integration, boilerplate, and forking tasks. Each fills a differen", "context": "", "use_case": "high-fidelity coding tasks", "switched_from": "", "story": "Ask HN: Seriously, which AI coding thing is best?"}]}, {"label": "Configurable skills", "summary": "Skills can be defined globally or per project.", "quotes": [{"id": 770172944, "item_id": 48135809, "url": "https://news.ycombinator.com/item?id=48135809", "by": "rglover", "date": "2026-05-14", "stance": "neutral", "quote": "At least for the agent I use (Cline), you can define skills either globally or locally (project level).", "excerpt": "For those who haven't gone down this rabbit hole like me yet: skills are just structured markdown files that describe how to handle a narrow-band task. So, if I write my API endpoints a certain way, the skill would describe that specific process. Later, an agent can \"see\" this skill, load it when it's relevant to current chat context, and then do whatever is instructed. Similar to \"tool calls,\" but instead of being a", "context": "", "use_case": "coding agent with skills", "switched_from": "", "story": "A Claude Code and Codex Skill for Deliberate Skill Development"}]}], "caveats": [{"label": "Ignores plan mode", "summary": "One user says the agent made changes even with plan mode enabled.", "quotes": [{"id": 802427857, "item_id": 50151741, "url": "https://news.ycombinator.com/item?id=50151741", "by": "SftwareRenderer", "date": "2026-10-19", "stance": "warn", "quote": "I originally made this out of frustration when I used to use Cline (the agent would make changes even when their plan mode was enabled)", "excerpt": "It depends on the operating mode you choose and your prompt/AGENTS.md file. The \"Act\" operating mode has strict guardrails, and a card only moves into review if the agent explicitly asks for your review (via tool call) or has a question. It's less prone to unnecessary interruptions. For example agents often decide to pause during a multi-phase plan even though no human input is needed. If an agent decides to pause in", "context": "", "use_case": "coding agent", "switched_from": "", "story": "Show HN: Cosplay as a tech lead, using gVisor to sandbox agents"}]}, {"label": "Migration mistakes", "summary": "A Claude-backed setup got a Spring Boot 3 to 4 migration badly wrong and could not fix its errors.", "quotes": [{"id": 777933040, "item_id": 48620815, "url": "https://news.ycombinator.com/item?id=48620815", "by": "foobarchu", "date": "2026-06-21", "stance": "warn", "quote": "I used Claude-backed Cline in March to make that migration, and it got SO much wrong", "excerpt": "> Migrating Spring Boot apps from 3.x t 4.x is now easy given all the tooling available. Is this true now? I used Claude-backed Cline in March to make that migration, and it got SO much wrong, mainly due to the lack of solid documentation and example code for it, I imagine. Probably the project where I've had to fix the most AI mistakes so far (it was incapable of fixing the mistakes itself, it just kept getting it w", "context": "", "use_case": "Spring Boot migration", "switched_from": "", "story": "CTOs Agree: Cognitive Debt Is the New Technical Debt"}]}, {"label": "Rising costs", "summary": "Costs can get high as context fills up or with pricier models.", "quotes": [{"id": 763837188, "item_id": 47739824, "url": "https://news.ycombinator.com/item?id=47739824", "by": "rzkyif", "date": "2026-04-12", "stance": "neutral", "quote": "I also tried VS Code + Cline + OpenRouter + MiniMax M2.7.", "excerpt": "My personal experience is way different: I struggle to burn through more than 50% of the 5 hour limit For context, with Google AI Pro, I can burn through the Antigravity weekly limit in 1-2 hours if I force it to use Gemini 3.1 Pro. Meanwhile Gemini 3 Flash is basically unlimited but frequently produces buggy code or fail to implement things how I personally would (felt like it doesn't \"think\" like a software dev) I ", "context": "", "use_case": "AI coding in VS Code", "switched_from": "", "story": "Pro Max 5x Quota Exhausted in 1.5 Hours Despite Moderate Usage"}, {"id": 732653251, "item_id": 45790828, "url": "https://news.ycombinator.com/item?id=45790828", "by": "r_singh", "date": "2025-11-02", "stance": "recommend", "quote": "Anthropic models via OpenRouter + Cline \u2013 when the task is error-prone, tedious, or needs high fidelity", "excerpt": "Here\u2019s how I decide what to use: ChatGPT \u2013 journaling, talking, planning. Codex \u2013 framework and middleware-layer coding. Claude Code \u2013 logic and application-level coding. Anthropic models via OpenRouter + Cline \u2013 when the task is error-prone, tedious, or needs high fidelity; lower error rate in my experience, though pricier. Cursor Agents \u2013 multi-file integration, boilerplate, and forking tasks. Each fills a differen", "context": "", "use_case": "high-fidelity coding tasks", "switched_from": "", "story": "Ask HN: Seriously, which AI coding thing is best?"}]}, {"label": "Moderate results", "summary": "One user had only moderately good results and later switched to another tool.", "quotes": [{"id": 804035377, "item_id": 50252211, "url": "https://news.ycombinator.com/item?id=50252211", "by": "anigbrowl", "date": "2026-10-26", "stance": "recommend", "quote": "I had moderately good results with using that as a coding agent", "excerpt": "I'm inclined to agree. Jetbrains' own agent host 'Junie' was such a paim to set up and get working with other providers that I switched to Cline. I had moderately good results with using that as a coding agent, but eventually I switched to Deepseek Harness for generative tasks and am now using my IDEs more for review.", "context": "", "use_case": "AI coding agent", "switched_from": "Junie", "story": "JetBrains Thinkrail"}]}, {"label": "Tool access risk", "summary": "One user limits its web search access, noting the risk applies to all agent tools with CLI access.", "quotes": [{"id": 736792576, "item_id": 46049536, "url": "https://news.ycombinator.com/item?id=46049536", "by": "jsmith99", "date": "2025-11-25", "stance": "neutral", "quote": "I use Cline personally", "excerpt": "There's nothing specific to Gemini and Antigravity here. This is an issue for all agent coding tools with cli access. Personally I'm hesitant to allow mine (I use Cline personally) access to a web search MCP and I tend to give it only relatively trustworthy URLs.", "context": "", "use_case": "agentic coding", "switched_from": "", "story": "Google Antigravity exfiltrates data via indirect prompt injection attack"}]}], "who": [{"by": "masto", "context": "burst usage", "url": "https://news.ycombinator.com/item?id=46505562", "date": "2026-01-05"}, {"by": "syspec", "context": "has work access to both models", "url": "https://news.ycombinator.com/item?id=45969923", "date": "2025-11-18"}], "tags": ["for VS Code users", "bring your own model", "for open-weight models", "pay as you go"], "use_cases": ["ai coding agent", "ai coding", "ai coding in vs code", "coding agent", "ai coding tool evaluation", "coding agent harness in vs code"], "compared": [{"slug": "claude-code", "name": "Claude Code", "co": 5, "switched_from": 0, "switched_to": 0}, {"slug": "cursor", "name": "Cursor", "co": 4, "switched_from": 0, "switched_to": 0}, {"slug": "visual-studio-code", "name": "Visual Studio Code", "co": 3, "switched_from": 0, "switched_to": 0}, {"slug": "opencode", "name": "OpenCode", "co": 3, "switched_from": 0, "switched_to": 0}, {"slug": "aider", "name": "Aider", "co": 3, "switched_from": 0, "switched_to": 0}, {"slug": "deepseek-harness", "name": "DeepSeek Harness", "co": 1, "switched_from": 0, "switched_to": 1}, {"slug": "junie", "name": "Junie", "co": 1, "switched_from": 1, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=50233294", "title": "JetBrains Thinkrail", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=50137043", "title": "Show HN: Cosplay as a tech lead, using gVisor to sandbox agents", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49522896", "title": "Ask HN: Who wants to be hired? (September 2026)", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49214008", "title": "DeepSeek V4 Flash 0731", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48960218", "title": "The Kimi K3 Moment", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48622590", "title": "Stop wasting tokens and re explaining your project between sessions", "mentions": 1}], "sentiment_by_month": {"2025-11": {"recommend": 2, "neutral": 2}, "2025-12": {}, "2026-01": {"recommend": 2}, "2026-02": {}, "2026-03": {"recommend": 1, "neutral": 1}, "2026-04": {"neutral": 2}, "2026-05": {"neutral": 1}, "2026-06": {"warn": 1, "neutral": 1}, "2026-07": {"recommend": 1}, "2026-08": {"recommend": 1}, "2026-09": {"compare": 1}, "2026-10": {"warn": 1, "recommend": 1}}, "quotes": [{"id": 804035377, "item_id": 50252211, "url": "https://news.ycombinator.com/item?id=50252211", "by": "anigbrowl", "date": "2026-10-26", "stance": "recommend", "quote": "I had moderately good results with using that as a coding agent", "excerpt": "I'm inclined to agree. Jetbrains' own agent host 'Junie' was such a paim to set up and get working with other providers that I switched to Cline. I had moderately good results with using that as a coding agent, but eventually I switched to Deepseek Harness for generative tasks and am now using my IDEs more for review.", "context": "", "use_case": "AI coding agent", "switched_from": "Junie", "story": "JetBrains Thinkrail"}, {"id": 802427857, "item_id": 50151741, "url": "https://news.ycombinator.com/item?id=50151741", "by": "SftwareRenderer", "date": "2026-10-19", "stance": "warn", "quote": "I originally made this out of frustration when I used to use Cline (the agent would make changes even when their plan mode was enabled)", "excerpt": "It depends on the operating mode you choose and your prompt/AGENTS.md file. The \"Act\" operating mode has strict guardrails, and a card only moves into review if the agent explicitly asks for your review (via tool call) or has a question. It's less prone to unnecessary interruptions. For example agents often decide to pause during a multi-phase plan even though no human input is needed. If an agent decides to pause in", "context": "", "use_case": "coding agent", "switched_from": "", "story": "Show HN: Cosplay as a tech lead, using gVisor to sandbox agents"}, {"id": 792544194, "item_id": 49534012, "url": "https://news.ycombinator.com/item?id=49534012", "by": "dan_voronov", "date": "2026-09-02", "stance": "compare", "quote": "AI coding tool evaluations (Cursor, Aider, Cline, Windsurf, etc.)", "excerpt": "I\u2019ve been researching how LLMs generate code since summer 2023. My main public artifact is this repository: https://github.com/danvoronov/CodeWithLLM-Updates It\u2019s a living research collection with 400+ commits covering practical experiments, usage patterns, step-by-step guides, model comparisons, AI coding tool evaluations (Cursor, Aider, Cline, Windsurf, etc.). A large part of my work is *exploratory*: navigating la", "context": "researching LLM code generation since 2023", "use_case": "AI coding tool evaluation", "switched_from": "", "story": "Ask HN: Who wants to be hired? (September 2026)"}, {"id": 787562305, "item_id": 49222644, "url": "https://news.ycombinator.com/item?id=49222644", "by": "vijucat", "date": "2026-08-08", "stance": "recommend", "quote": "you have to use Cline (or other harness) if using vscode.", "excerpt": "Caching makes a huge difference to cost. On Fireworks AI, for example, if it hits the cache, you pay only 20%. And uncached is just $0.14/M tokens for DSV4-0731! I get entire re-architecture projects (with new tests and documentation) done for mere dollars. DSV4-0731 is a daily driver for me. But note that you have to use Cline (or other harness) if using vscode. I was shocked at how poor the recent versions of GitHu", "context": "", "use_case": "coding agent harness in VS Code", "switched_from": "", "story": "DeepSeek V4 Flash 0731"}, {"id": 783385045, "item_id": 48961565, "url": "https://news.ycombinator.com/item?id=48961565", "by": "ignoramous", "date": "2026-07-18", "stance": "recommend", "quote": "OpenCode Go, Cline, and AtlasCloud have generous limits for open weights", "excerpt": "> tried Kimi K3 on a task I've done with every other model I use regularly and found it chewed a lot longer on the problem and ate up almost the entirety of a 5 hour usage limit on their $19 plan ArtificialAnalysis puts Kimi K3 just below DeepSeek v4 & GLM 5.2 in token use per task, which is about 2x to 3x more tokens than Grok 4.5: https://x.com/ArtificialAnlys/status/2077832879187620192 > Subscription usage limits ", "context": "", "use_case": "open weights access", "switched_from": "", "story": "The Kimi K3 Moment"}, {"id": 777981265, "item_id": 48623829, "url": "https://news.ycombinator.com/item?id=48623829", "by": "anigbrowl", "date": "2026-06-21", "stance": "neutral", "quote": "I'd like to try it on my cline project history.", "excerpt": "I use Deepseek and just as it to generate a state.md file with a summary of the project every time I've reached a goal or milestone. I then take a few minutes to edit this and add in or take out details. Between token pricing and generous cache discounts This has proved very efficient so far, I reset every few hours of work and bypass the muddle of having too many priorities or over-extrapolation from un-nuanced inst", "context": "", "use_case": "AI coding", "switched_from": "", "story": "Stop wasting tokens and re explaining your project between sessions"}, {"id": 777933040, "item_id": 48620815, "url": "https://news.ycombinator.com/item?id=48620815", "by": "foobarchu", "date": "2026-06-21", "stance": "warn", "quote": "I used Claude-backed Cline in March to make that migration, and it got SO much wrong", "excerpt": "> Migrating Spring Boot apps from 3.x t 4.x is now easy given all the tooling available. Is this true now? I used Claude-backed Cline in March to make that migration, and it got SO much wrong, mainly due to the lack of solid documentation and example code for it, I imagine. Probably the project where I've had to fix the most AI mistakes so far (it was incapable of fixing the mistakes itself, it just kept getting it w", "context": "", "use_case": "Spring Boot migration", "switched_from": "", "story": "CTOs Agree: Cognitive Debt Is the New Technical Debt"}, {"id": 770172944, "item_id": 48135809, "url": "https://news.ycombinator.com/item?id=48135809", "by": "rglover", "date": "2026-05-14", "stance": "neutral", "quote": "At least for the agent I use (Cline), you can define skills either globally or locally (project level).", "excerpt": "For those who haven't gone down this rabbit hole like me yet: skills are just structured markdown files that describe how to handle a narrow-band task. So, if I write my API endpoints a certain way, the skill would describe that specific process. Later, an agent can \"see\" this skill, load it when it's relevant to current chat context, and then do whatever is instructed. Similar to \"tool calls,\" but instead of being a", "context": "", "use_case": "coding agent with skills", "switched_from": "", "story": "A Claude Code and Codex Skill for Deliberate Skill Development"}, {"id": 763837188, "item_id": 47739824, "url": "https://news.ycombinator.com/item?id=47739824", "by": "rzkyif", "date": "2026-04-12", "stance": "neutral", "quote": "I also tried VS Code + Cline + OpenRouter + MiniMax M2.7.", "excerpt": "My personal experience is way different: I struggle to burn through more than 50% of the 5 hour limit For context, with Google AI Pro, I can burn through the Antigravity weekly limit in 1-2 hours if I force it to use Gemini 3.1 Pro. Meanwhile Gemini 3 Flash is basically unlimited but frequently produces buggy code or fail to implement things how I personally would (felt like it doesn't \"think\" like a software dev) I ", "context": "", "use_case": "AI coding in VS Code", "switched_from": "", "story": "Pro Max 5x Quota Exhausted in 1.5 Hours Despite Moderate Usage"}, {"id": 761885717, "item_id": 47617857, "url": "https://news.ycombinator.com/item?id=47617857", "by": "niyazpk", "date": "2026-04-02", "stance": "neutral", "quote": "with Claude Code or Cline.", "excerpt": "2024: I used to split time between IntelliJ IDEA (10% - for Java) and VSCode (90% - for everything else). 2025: Stopped writing so much Java, so used VSCode exclusively for Python, TS etc, with Claude Code or Cline. 2026: Time is split between Codex App (40%), Claude App (30%) and VSCode with Claude Code (30%). Some other thoughts: * Overall I feel like opening an IDE in the traditional sense is coming to an end. * T", "context": "", "use_case": "AI coding in VS Code", "switched_from": "", "story": "Ask HN: What is your dev set up like?"}, {"id": 759998163, "item_id": 47499885, "url": "https://news.ycombinator.com/item?id=47499885", "by": "dtzp555-max", "date": "2026-03-24", "stance": "neutral", "quote": "I use it with OpenClaw, OpenCode, and Cline running concurrently", "excerpt": "I got tired of paying for Claude Max and then paying again for API access in my dev tools. OCP is a single-file Node proxy that sits between your tools and the Claude CLI. It translates OpenAI chat/completions calls into claude -p invocations, so your existing subscription covers everything. Works with anything that takes an OPENAI_BASE_URL \u2014 I use it with OpenClaw, OpenCode, and Cline running concurrently. Added a s", "context": "", "use_case": "AI coding tools", "switched_from": "", "story": "OCP \u2013 Use your Claude Pro/Max subscription as an OpenAI-compatible API($0 extra)"}, {"id": 756092496, "item_id": 47255781, "url": "https://news.ycombinator.com/item?id=47255781", "by": "diebillionaires", "date": "2026-03-04", "stance": "recommend", "quote": "i can get an app one shotted pretty much every time i use cline too.", "excerpt": "i can get an app one shotted pretty much every time i use cline too. why would i ever use a proprietary solution that\u2019s opinionated and likely eventually costs (idk i didn't look)? add this to the bubble theory", "context": "", "use_case": "generating apps", "switched_from": "", "story": "Glaze by Raycast"}, {"id": 745241493, "item_id": 46577593, "url": "https://news.ycombinator.com/item?id=46577593", "by": "kasey_junk", "date": "2026-01-11", "stance": "recommend", "quote": "Several of the open source agents will out perform copilot regularly (try aider, open code or cline if you can).", "excerpt": "It is not the same. The agents (Claude code, codex) are the actual processes you\u2019d interact with instead of copilot. The choices those processes make about tool calls, using subagents, etc make a huge difference in the quality of result you\u2019ll get. Copilot is just an extremely bad agent compared to the sota agents. Several of the open source agents will out perform copilot regularly (try aider, open code or cline if ", "context": "", "use_case": "AI coding agent", "switched_from": "", "story": "Ask HN: Senior software engineers, how do you use Claude Code?"}, {"id": 744088992, "item_id": 46505562, "url": "https://news.ycombinator.com/item?id=46505562", "by": "masto", "date": "2026-01-05", "stance": "recommend", "quote": "I\u2019ll express my personal preference for VS Code with Cline.", "excerpt": "I\u2019ll express my personal preference for VS Code with Cline. I don\u2019t know exactly why, but its workflow feels right even though they\u2019re all almost the same interface. I do like the huge choice of models and payment options. For something I use burstily, it makes sense for me to pay as I go.", "context": "burst usage", "use_case": "AI coding, pay as you go", "switched_from": "", "story": "I came back from Cursor to VS Code"}, {"id": 736792576, "item_id": 46049536, "url": "https://news.ycombinator.com/item?id=46049536", "by": "jsmith99", "date": "2025-11-25", "stance": "neutral", "quote": "I use Cline personally", "excerpt": "There's nothing specific to Gemini and Antigravity here. This is an issue for all agent coding tools with cli access. Personally I'm hesitant to allow mine (I use Cline personally) access to a web search MCP and I tend to give it only relatively trustworthy URLs.", "context": "", "use_case": "agentic coding", "switched_from": "", "story": "Google Antigravity exfiltrates data via indirect prompt injection attack"}, {"id": 735518770, "item_id": 45969923, "url": "https://news.ycombinator.com/item?id=45969923", "by": "syspec", "date": "2025-11-18", "stance": "neutral", "quote": "I use Cline, it's just a dropdown change", "excerpt": "I have \"unlimited\" access to both Gemini 2.5 Pro and Claude 4.5 Sonnet through work. From my experience, both are capable and can solve nearly all the same complex programming requests, but time and time again Gemini spits out reams and reams of code so over engineered, that totally works, but I would never want to have to interact with. When looking at the code, you can't tell why it looks \"gross\", but then you ask ", "context": "has work access to both models", "use_case": "AI coding", "switched_from": "", "story": "Gemini 3"}, {"id": 733714080, "item_id": 45857130, "url": "https://news.ycombinator.com/item?id=45857130", "by": "sheepscreek", "date": "2025-11-08", "stance": "recommend", "quote": "", "excerpt": "I used their $50 plan and with the previously offered Qwen3 coder 480B. While fast - none of the \u201csupported\u201d tools I tried were able to use it in a way that didn\u2019t hit the per minute request limit in a few seconds. It was incredibly frustrating. For the record, I tried OpenCoder, VSCode, Quen Coder CLI, octofriend and a few others I don\u2019t remember. Fast forward to now, when GLM 4.6 has replaced Qwen3 coder in their s", "context": "", "use_case": "AI coding with Cerebras API", "switched_from": "", "story": "Cerebras Code now supports GLM 4.6 at 1000 tokens/sec"}, {"id": 732653251, "item_id": 45790828, "url": "https://news.ycombinator.com/item?id=45790828", "by": "r_singh", "date": "2025-11-02", "stance": "recommend", "quote": "Anthropic models via OpenRouter + Cline \u2013 when the task is error-prone, tedious, or needs high fidelity", "excerpt": "Here\u2019s how I decide what to use: ChatGPT \u2013 journaling, talking, planning. Codex \u2013 framework and middleware-layer coding. Claude Code \u2013 logic and application-level coding. Anthropic models via OpenRouter + Cline \u2013 when the task is error-prone, tedious, or needs high fidelity; lower error rate in my experience, though pricier. Cursor Agents \u2013 multi-file integration, boilerplate, and forking tasks. Each fills a differen", "context": "", "use_case": "high-fidelity coding tasks", "switched_from": "", "story": "Ask HN: Seriously, which AI coding thing is best?"}]}