{"slug": "caveman", "name": "caveman", "category": "ai-coding", "category_name": "AI coding tools", "description": "A skill for AI coding agents that makes them reply in terse, caveman-style language.", "recommenders": 12, "commenters": 16, "warners": 1, "score": 11.322, "rising": 6.69, "first": "2026-05-02", "latest": "2026-10-29", "why": "Liked for making coding agents give short, matter-of-fact answers instead of verbose output, with token savings seen as a secondary bonus.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 12, "neutral": 3, "compare": 0, "warn": 1}, "spark": [0, 0, 0, 0, 0, 0, 1, 0, 3, 3, 4, 1], "listed": true, "loved": ["Terse responses", "Token savings"], "caveat": "Modest effect", "short_quote": "a coworker told me about Caveman and it has made my Claude experience vastly better.", "icon": true, "domain": null, "github_repo": "github.com/juliusbrussee/caveman", "official_url": "https://github.com/JuliusBrussee/caveman", "pricing": "unknown", "aliases": ["caveman", "Caveman", "caveman skill"], "verdict": [{"text": "Liked for making coding agents give short, matter-of-fact answers instead of verbose output", "cites": [{"id": 794666064, "item_id": 49666629, "url": "https://news.ycombinator.com/item?id=49666629", "by": "theshrike79", "date": "2026-09-11", "stance": "recommend", "quote": "I use it to make the agent shut up.", "excerpt": "You use caveman to save on tokens, I use it to make the agent shut up. We are not the same =) When caveman was released, token efficiency was a lot more relevant. But in the time I've actually come to like its matter-of-fact and short way of responding. There are other similar ones, but caveman is familiar.", "context": "", "use_case": "terse agent responses", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}, {"id": 794324448, "item_id": 49645278, "url": "https://news.ycombinator.com/item?id=49645278", "by": "eleventen", "date": "2026-09-10", "stance": "recommend", "quote": "Same. I use this to fix claudespeak.", "excerpt": "Same. I use this to fix claudespeak. Any hypothetical token savings are just gravy. https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/SKILL.md", "context": "", "use_case": "terse LLM responses", "switched_from": "", "story": "DeepSeek v4.1 Flash"}, {"id": 790057056, "item_id": 49378566, "url": "https://news.ycombinator.com/item?id=49378566", "by": "ARandomerDude", "date": "2026-08-20", "stance": "recommend", "quote": "a coworker told me about Caveman and it has made my Claude experience vastly better.", "excerpt": "Along similar lines, a coworker told me about Caveman and it has made my Claude experience vastly better. I'm already a programmer, Claude shouldn't need to tell me like I'm 5. https://github.com/JuliusBrussee/caveman", "context": "programmer", "use_case": "terse Claude responses", "switched_from": "", "story": "Clean up Claude 5's token vomit with a separate LLM"}, {"id": 788844177, "item_id": 49302761, "url": "https://news.ycombinator.com/item?id=49302761", "by": "mrinterweb", "date": "2026-08-14", "stance": "recommend", "quote": "I use the caveman skill, and I think that does help some.", "excerpt": "The pain points in the article do not bother me. I'm bothered by Opus 5's verbosity. It is so long-winded and you have to read through verbose outputs to mentally distill what is important. It is exhausting. I don't think I've ever started skim reading LLM output more that I do with Opus 5. I use the caveman skill, and I think that does help some.", "context": "", "use_case": "reducing verbosity", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}, {"id": 782488896, "item_id": 48905556, "url": "https://news.ycombinator.com/item?id=48905556", "by": "bunderbunder", "date": "2026-07-14", "stance": "recommend", "quote": "I recently started using caveman, and it\u2019s been great.", "excerpt": "I recently started using caveman, and it\u2019s been great. It doesn\u2019t just cut down on overuse of specific terms; it cuts down on time spent digesting slop in general. https://github.com/JuliusBrussee/caveman", "context": "", "use_case": "cutting verbose AI output", "switched_from": "", "story": "How to stop Claude from saying load-bearing"}]}, {"text": ", with token savings seen as a secondary bonus.", "cites": [{"id": 804718304, "item_id": 50294894, "url": "https://news.ycombinator.com/item?id=50294894", "by": "d3Xt3r", "date": "2026-10-29", "stance": "recommend", "quote": "In fact, there's even a skill for it: https://github.com/juliusbrussee/caveman", "excerpt": "Neither, just be neutral and use caveman speak. It saves tokens. Being polite consumes extra tokens, and being rude can risk you getting banned. So use caveman speak, save tokens. The best part is, you'll get curt responses as well, without any cringy flowery language that LLMs are known for. In fact, there's even a skill for it: https://github.com/juliusbrussee/caveman", "context": "", "use_case": "saving tokens with agents", "switched_from": "", "story": "Ask HN: Should I be polite to my agent or should I treat it like shit?"}, {"id": 794324448, "item_id": 49645278, "url": "https://news.ycombinator.com/item?id=49645278", "by": "eleventen", "date": "2026-09-10", "stance": "recommend", "quote": "Same. I use this to fix claudespeak.", "excerpt": "Same. I use this to fix claudespeak. Any hypothetical token savings are just gravy. https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/SKILL.md", "context": "", "use_case": "terse LLM responses", "switched_from": "", "story": "DeepSeek v4.1 Flash"}, {"id": 794666064, "item_id": 49666629, "url": "https://news.ycombinator.com/item?id=49666629", "by": "theshrike79", "date": "2026-09-11", "stance": "recommend", "quote": "I use it to make the agent shut up.", "excerpt": "You use caveman to save on tokens, I use it to make the agent shut up. We are not the same =) When caveman was released, token efficiency was a lot more relevant. But in the time I've actually come to like its matter-of-fact and short way of responding. There are other similar ones, but caveman is familiar.", "context": "", "use_case": "terse agent responses", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}]}, {"text": " The most common caveat is that its effect is modest or doubtful", "cites": [{"id": 794576674, "item_id": 49661042, "url": "https://news.ycombinator.com/item?id=49661042", "by": "kriskrunch", "date": "2026-09-11", "stance": "recommend", "quote": "Ponytail and Caveman seem to help somewhat.", "excerpt": "Open question, how does this instruction in agent-rules.md look? \"Cap large/unknown command output: `COMMAND 2>&1 | head -c 4000`. Never stream full logs, tests, or large files.\" I use that instead of RTK. Empirically, I found RTK makes my agents run longer to complete similar tasks. Ponytail and Caveman seem to help somewhat.", "context": "", "use_case": "agent token savings", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}, {"id": 788844177, "item_id": 49302761, "url": "https://news.ycombinator.com/item?id=49302761", "by": "mrinterweb", "date": "2026-08-14", "stance": "recommend", "quote": "I use the caveman skill, and I think that does help some.", "excerpt": "The pain points in the article do not bother me. I'm bothered by Opus 5's verbosity. It is so long-winded and you have to read through verbose outputs to mentally distill what is important. It is exhausting. I don't think I've ever started skim reading LLM output more that I do with Opus 5. I use the caveman skill, and I think that does help some.", "context": "", "use_case": "reducing verbosity", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}, {"id": 794505953, "item_id": 49656622, "url": "https://news.ycombinator.com/item?id=49656622", "by": "aeneas_ory", "date": "2026-09-11", "stance": "warn", "quote": "Whether it's caveman, RTK, or whatever other vibe-coded productivity/token cost saving hacks/skills/claude.md.", "excerpt": "All of these \"hacks\" are snakeoil and I think deep down we all know. Whether it's caveman, RTK, or whatever other vibe-coded productivity/token cost saving hacks/skills/claude.md. What I had success with (although benchmarks are older) is to index the codebase with a dedicated local code embedding model. It's a bit expensive on the CPU side but in my benchmarks it reduced token use and wall clock time significantly. ", "context": "", "use_case": "token savings", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}, {"id": 800071153, "item_id": 50004447, "url": "https://news.ycombinator.com/item?id=50004447", "by": "jeremyjh", "date": "2026-10-08", "stance": "neutral", "quote": "I don't use any plugins or skill libraries apart from Caveman and I'm not sure how useful that really is anymore", "excerpt": "I use oh-my-pi (omp.sh) - mostly with stock settings and skills but its a \"fully loaded\" harness that you don't really have to add anything to. I do change a couple of things: I set it to prefer subagents, and I enable rewind. It is crazy that rewind is not a default, it saves a TON of context - if the agent goes down a crazy path that burns a lot of tokens it can rewind to an earlier checkpoint with the exploration ", "context": "", "use_case": "reducing context usage", "switched_from": "", "story": "Claude Haiku 5.5"}]}, {"text": ", and one user says the model starts ignoring it after a while.", "cites": [{"id": 790362339, "item_id": 49397646, "url": "https://news.ycombinator.com/item?id=49397646", "by": "fractalf", "date": "2026-08-22", "stance": "neutral", "quote": "I use the caveman skill, but after a while it is just ignored", "excerpt": "I usually dont comment, but just had to. Opus 5 talls garbage, Im back om 4.8 which is better but still verbose. I use the caveman skill, but after a while it is just ignored.. I switched to Deepseek for a task lately. It was a wow experience (the language part)", "context": "", "use_case": "reducing verbosity", "switched_from": "", "story": "Claudette: Make Claude Stop Talking Like a BuzzFeed Article"}]}], "praise": [{"label": "Terse responses", "summary": "Commenters use it to cut LLM verbosity and get short, matter-of-fact replies.", "quotes": [{"id": 794666064, "item_id": 49666629, "url": "https://news.ycombinator.com/item?id=49666629", "by": "theshrike79", "date": "2026-09-11", "stance": "recommend", "quote": "I use it to make the agent shut up.", "excerpt": "You use caveman to save on tokens, I use it to make the agent shut up. We are not the same =) When caveman was released, token efficiency was a lot more relevant. But in the time I've actually come to like its matter-of-fact and short way of responding. There are other similar ones, but caveman is familiar.", "context": "", "use_case": "terse agent responses", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}, {"id": 790057056, "item_id": 49378566, "url": "https://news.ycombinator.com/item?id=49378566", "by": "ARandomerDude", "date": "2026-08-20", "stance": "recommend", "quote": "a coworker told me about Caveman and it has made my Claude experience vastly better.", "excerpt": "Along similar lines, a coworker told me about Caveman and it has made my Claude experience vastly better. I'm already a programmer, Claude shouldn't need to tell me like I'm 5. https://github.com/JuliusBrussee/caveman", "context": "programmer", "use_case": "terse Claude responses", "switched_from": "", "story": "Clean up Claude 5's token vomit with a separate LLM"}, {"id": 782488896, "item_id": 48905556, "url": "https://news.ycombinator.com/item?id=48905556", "by": "bunderbunder", "date": "2026-07-14", "stance": "recommend", "quote": "I recently started using caveman, and it\u2019s been great.", "excerpt": "I recently started using caveman, and it\u2019s been great. It doesn\u2019t just cut down on overuse of specific terms; it cuts down on time spent digesting slop in general. https://github.com/JuliusBrussee/caveman", "context": "", "use_case": "cutting verbose AI output", "switched_from": "", "story": "How to stop Claude from saying load-bearing"}]}, {"label": "Token savings", "summary": "Some use it to save tokens, though others treat the savings as a side benefit.", "quotes": [{"id": 804718304, "item_id": 50294894, "url": "https://news.ycombinator.com/item?id=50294894", "by": "d3Xt3r", "date": "2026-10-29", "stance": "recommend", "quote": "In fact, there's even a skill for it: https://github.com/juliusbrussee/caveman", "excerpt": "Neither, just be neutral and use caveman speak. It saves tokens. Being polite consumes extra tokens, and being rude can risk you getting banned. So use caveman speak, save tokens. The best part is, you'll get curt responses as well, without any cringy flowery language that LLMs are known for. In fact, there's even a skill for it: https://github.com/juliusbrussee/caveman", "context": "", "use_case": "saving tokens with agents", "switched_from": "", "story": "Ask HN: Should I be polite to my agent or should I treat it like shit?"}, {"id": 794324448, "item_id": 49645278, "url": "https://news.ycombinator.com/item?id=49645278", "by": "eleventen", "date": "2026-09-10", "stance": "recommend", "quote": "Same. I use this to fix claudespeak.", "excerpt": "Same. I use this to fix claudespeak. Any hypothetical token savings are just gravy. https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/SKILL.md", "context": "", "use_case": "terse LLM responses", "switched_from": "", "story": "DeepSeek v4.1 Flash"}, {"id": 781440880, "item_id": 48840055, "url": "https://news.ycombinator.com/item?id=48840055", "by": "hoppp", "date": "2026-07-08", "stance": "recommend", "quote": "It's for getting it to output shorter answers, but also could help with your burnout.", "excerpt": "Tell the llm to answer like a cavemen, if llm talk like cavemen, the answers become shorter and more compressed. https://github.com/JuliusBrussee/caveman It's for getting it to output shorter answers, but also could help with your burnout.", "context": "", "use_case": "shorter LLM answers", "switched_from": "", "story": "I Think I Have LLM Burnout"}]}, {"label": "Less reading", "summary": "Users say it cuts the time spent reading and digesting long model output.", "quotes": [{"id": 782488896, "item_id": 48905556, "url": "https://news.ycombinator.com/item?id=48905556", "by": "bunderbunder", "date": "2026-07-14", "stance": "recommend", "quote": "I recently started using caveman, and it\u2019s been great.", "excerpt": "I recently started using caveman, and it\u2019s been great. It doesn\u2019t just cut down on overuse of specific terms; it cuts down on time spent digesting slop in general. https://github.com/JuliusBrussee/caveman", "context": "", "use_case": "cutting verbose AI output", "switched_from": "", "story": "How to stop Claude from saying load-bearing"}, {"id": 788844177, "item_id": 49302761, "url": "https://news.ycombinator.com/item?id=49302761", "by": "mrinterweb", "date": "2026-08-14", "stance": "recommend", "quote": "I use the caveman skill, and I think that does help some.", "excerpt": "The pain points in the article do not bother me. I'm bothered by Opus 5's verbosity. It is so long-winded and you have to read through verbose outputs to mentally distill what is important. It is exhausting. I don't think I've ever started skim reading LLM output more that I do with Opus 5. I use the caveman skill, and I think that does help some.", "context": "", "use_case": "reducing verbosity", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}]}], "caveats": [{"label": "Modest effect", "summary": "Even users who recommend it say it helps only somewhat.", "quotes": [{"id": 794576674, "item_id": 49661042, "url": "https://news.ycombinator.com/item?id=49661042", "by": "kriskrunch", "date": "2026-09-11", "stance": "recommend", "quote": "Ponytail and Caveman seem to help somewhat.", "excerpt": "Open question, how does this instruction in agent-rules.md look? \"Cap large/unknown command output: `COMMAND 2>&1 | head -c 4000`. Never stream full logs, tests, or large files.\" I use that instead of RTK. Empirically, I found RTK makes my agents run longer to complete similar tasks. Ponytail and Caveman seem to help somewhat.", "context": "", "use_case": "agent token savings", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}, {"id": 788844177, "item_id": 49302761, "url": "https://news.ycombinator.com/item?id=49302761", "by": "mrinterweb", "date": "2026-08-14", "stance": "recommend", "quote": "I use the caveman skill, and I think that does help some.", "excerpt": "The pain points in the article do not bother me. I'm bothered by Opus 5's verbosity. It is so long-winded and you have to read through verbose outputs to mentally distill what is important. It is exhausting. I don't think I've ever started skim reading LLM output more that I do with Opus 5. I use the caveman skill, and I think that does help some.", "context": "", "use_case": "reducing verbosity", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}]}, {"label": "Doubtful value", "summary": "One commenter calls such token-saving skills snake oil, and another is unsure how useful it still is.", "quotes": [{"id": 794505953, "item_id": 49656622, "url": "https://news.ycombinator.com/item?id=49656622", "by": "aeneas_ory", "date": "2026-09-11", "stance": "warn", "quote": "Whether it's caveman, RTK, or whatever other vibe-coded productivity/token cost saving hacks/skills/claude.md.", "excerpt": "All of these \"hacks\" are snakeoil and I think deep down we all know. Whether it's caveman, RTK, or whatever other vibe-coded productivity/token cost saving hacks/skills/claude.md. What I had success with (although benchmarks are older) is to index the codebase with a dedicated local code embedding model. It's a bit expensive on the CPU side but in my benchmarks it reduced token use and wall clock time significantly. ", "context": "", "use_case": "token savings", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}, {"id": 800071153, "item_id": 50004447, "url": "https://news.ycombinator.com/item?id=50004447", "by": "jeremyjh", "date": "2026-10-08", "stance": "neutral", "quote": "I don't use any plugins or skill libraries apart from Caveman and I'm not sure how useful that really is anymore", "excerpt": "I use oh-my-pi (omp.sh) - mostly with stock settings and skills but its a \"fully loaded\" harness that you don't really have to add anything to. I do change a couple of things: I set it to prefer subagents, and I enable rewind. It is crazy that rewind is not a default, it saves a TON of context - if the agent goes down a crazy path that burns a lot of tokens it can rewind to an earlier checkpoint with the exploration ", "context": "", "use_case": "reducing context usage", "switched_from": "", "story": "Claude Haiku 5.5"}]}, {"label": "Gets ignored", "summary": "One user reports the model stops following the skill after a while.", "quotes": [{"id": 790362339, "item_id": 49397646, "url": "https://news.ycombinator.com/item?id=49397646", "by": "fractalf", "date": "2026-08-22", "stance": "neutral", "quote": "I use the caveman skill, but after a while it is just ignored", "excerpt": "I usually dont comment, but just had to. Opus 5 talls garbage, Im back om 4.8 which is better but still verbose. I use the caveman skill, but after a while it is just ignored.. I switched to Deepseek for a task lately. It was a wow experience (the language part)", "context": "", "use_case": "reducing verbosity", "switched_from": "", "story": "Claudette: Make Claude Stop Talking Like a BuzzFeed Article"}]}, {"label": "Savings less relevant", "summary": "One user says token efficiency matters less now than when it was released.", "quotes": [{"id": 794666064, "item_id": 49666629, "url": "https://news.ycombinator.com/item?id=49666629", "by": "theshrike79", "date": "2026-09-11", "stance": "recommend", "quote": "I use it to make the agent shut up.", "excerpt": "You use caveman to save on tokens, I use it to make the agent shut up. We are not the same =) When caveman was released, token efficiency was a lot more relevant. But in the time I've actually come to like its matter-of-fact and short way of responding. There are other similar ones, but caveman is familiar.", "context": "", "use_case": "terse agent responses", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}, {"id": 794324448, "item_id": 49645278, "url": "https://news.ycombinator.com/item?id=49645278", "by": "eleventen", "date": "2026-09-10", "stance": "recommend", "quote": "Same. I use this to fix claudespeak.", "excerpt": "Same. I use this to fix claudespeak. Any hypothetical token savings are just gravy. https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/SKILL.md", "context": "", "use_case": "terse LLM responses", "switched_from": "", "story": "DeepSeek v4.1 Flash"}]}], "who": [{"by": "ARandomerDude", "context": "programmer", "url": "https://news.ycombinator.com/item?id=49378566", "date": "2026-08-20"}], "tags": ["for Claude users", "agent skill", "for experienced programmers"], "use_cases": ["reducing verbosity", "saving tokens with agents", "reducing context usage", "terse agent responses", "agent token savings", "token savings"], "compared": [{"slug": "rtk", "name": "RTK", "co": 3, "switched_from": 0, "switched_to": 0}, {"slug": "claude-opus", "name": "Claude Opus", "co": 2, "switched_from": 0, "switched_to": 0}], "makers": [{"id": 771947299, "item_id": 48246706, "url": "https://news.ycombinator.com/item?id=48246706", "by": "phillc73", "date": "2026-05-23", "stance": "recommend", "quote": "this skill removes all extraneous response text, including articles.", "excerpt": "I was a Mistral Le Chat Pro subscriber (the \u20ac20/month plan). Yesterday I hit my monthly limit. Switching to PAYG I burned through another \u20ac40 in one evening, working on the same project, with the same tasks. I upgraded my plan last night to Mistral Le Chat Teams. This now costs me \u20ac60 per month for two users. Limits have been reset, but I have no idea now if my per seat limit is higher than the Pro plan, or if the li", "context": "", "use_case": "reducing output token usage", "switched_from": "", "story": "Microsoft Drops Claude Code After Budget Overrun"}], "threads": [{"url": "https://news.ycombinator.com/item?id=49656471", "title": "RTK reports token savings, but our cost benchmarks disagree", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=50290514", "title": "Ask HN: Should I be polite to my agent or should I treat it like shit?", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49996437", "title": "Claude Haiku 5.5", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49639090", "title": "DeepSeek v4.1 Flash", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49571465", "title": "Portal by Spotify cut my Claude Code token usage by 90%", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49388752", "title": "Claudette: Make Claude Stop Talking Like a BuzzFeed Article", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {}, "2026-03": {}, "2026-04": {}, "2026-05": {"recommend": 1}, "2026-06": {}, "2026-07": {"recommend": 3}, "2026-08": {"recommend": 3, "neutral": 2}, "2026-09": {"recommend": 4, "warn": 1}, "2026-10": {"neutral": 1, "recommend": 1}}, "quotes": [{"id": 804718304, "item_id": 50294894, "url": "https://news.ycombinator.com/item?id=50294894", "by": "d3Xt3r", "date": "2026-10-29", "stance": "recommend", "quote": "In fact, there's even a skill for it: https://github.com/juliusbrussee/caveman", "excerpt": "Neither, just be neutral and use caveman speak. It saves tokens. Being polite consumes extra tokens, and being rude can risk you getting banned. So use caveman speak, save tokens. The best part is, you'll get curt responses as well, without any cringy flowery language that LLMs are known for. In fact, there's even a skill for it: https://github.com/juliusbrussee/caveman", "context": "", "use_case": "saving tokens with agents", "switched_from": "", "story": "Ask HN: Should I be polite to my agent or should I treat it like shit?"}, {"id": 800071153, "item_id": 50004447, "url": "https://news.ycombinator.com/item?id=50004447", "by": "jeremyjh", "date": "2026-10-08", "stance": "neutral", "quote": "I don't use any plugins or skill libraries apart from Caveman and I'm not sure how useful that really is anymore", "excerpt": "I use oh-my-pi (omp.sh) - mostly with stock settings and skills but its a \"fully loaded\" harness that you don't really have to add anything to. I do change a couple of things: I set it to prefer subagents, and I enable rewind. It is crazy that rewind is not a default, it saves a TON of context - if the agent goes down a crazy path that burns a lot of tokens it can rewind to an earlier checkpoint with the exploration ", "context": "", "use_case": "reducing context usage", "switched_from": "", "story": "Claude Haiku 5.5"}, {"id": 794666064, "item_id": 49666629, "url": "https://news.ycombinator.com/item?id=49666629", "by": "theshrike79", "date": "2026-09-11", "stance": "recommend", "quote": "I use it to make the agent shut up.", "excerpt": "You use caveman to save on tokens, I use it to make the agent shut up. We are not the same =) When caveman was released, token efficiency was a lot more relevant. But in the time I've actually come to like its matter-of-fact and short way of responding. There are other similar ones, but caveman is familiar.", "context": "", "use_case": "terse agent responses", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}, {"id": 794576674, "item_id": 49661042, "url": "https://news.ycombinator.com/item?id=49661042", "by": "kriskrunch", "date": "2026-09-11", "stance": "recommend", "quote": "Ponytail and Caveman seem to help somewhat.", "excerpt": "Open question, how does this instruction in agent-rules.md look? \"Cap large/unknown command output: `COMMAND 2>&1 | head -c 4000`. Never stream full logs, tests, or large files.\" I use that instead of RTK. Empirically, I found RTK makes my agents run longer to complete similar tasks. Ponytail and Caveman seem to help somewhat.", "context": "", "use_case": "agent token savings", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}, {"id": 794505953, "item_id": 49656622, "url": "https://news.ycombinator.com/item?id=49656622", "by": "aeneas_ory", "date": "2026-09-11", "stance": "warn", "quote": "Whether it's caveman, RTK, or whatever other vibe-coded productivity/token cost saving hacks/skills/claude.md.", "excerpt": "All of these \"hacks\" are snakeoil and I think deep down we all know. Whether it's caveman, RTK, or whatever other vibe-coded productivity/token cost saving hacks/skills/claude.md. What I had success with (although benchmarks are older) is to index the codebase with a dedicated local code embedding model. It's a bit expensive on the CPU side but in my benchmarks it reduced token use and wall clock time significantly. ", "context": "", "use_case": "token savings", "switched_from": "", "story": "RTK reports token savings, but our cost benchmarks disagree"}, {"id": 794324448, "item_id": 49645278, "url": "https://news.ycombinator.com/item?id=49645278", "by": "eleventen", "date": "2026-09-10", "stance": "recommend", "quote": "Same. I use this to fix claudespeak.", "excerpt": "Same. I use this to fix claudespeak. Any hypothetical token savings are just gravy. https://github.com/JuliusBrussee/caveman/blob/main/skills/caveman/SKILL.md", "context": "", "use_case": "terse LLM responses", "switched_from": "", "story": "DeepSeek v4.1 Flash"}, {"id": 793489968, "item_id": 49593123, "url": "https://news.ycombinator.com/item?id=49593123", "by": "nerdyadventurer", "date": "2026-09-07", "stance": "recommend", "quote": "There is also Caveman[1] skill which rewrite prompts.", "excerpt": "There is also Caveman[1] skill which rewrite prompts. 1. https://github.com/JuliusBrussee/caveman/", "context": "", "use_case": "rewriting prompts to save tokens", "switched_from": "", "story": "Portal by Spotify cut my Claude Code token usage by 90%"}, {"id": 790362339, "item_id": 49397646, "url": "https://news.ycombinator.com/item?id=49397646", "by": "fractalf", "date": "2026-08-22", "stance": "neutral", "quote": "I use the caveman skill, but after a while it is just ignored", "excerpt": "I usually dont comment, but just had to. Opus 5 talls garbage, Im back om 4.8 which is better but still verbose. I use the caveman skill, but after a while it is just ignored.. I switched to Deepseek for a task lately. It was a wow experience (the language part)", "context": "", "use_case": "reducing verbosity", "switched_from": "", "story": "Claudette: Make Claude Stop Talking Like a BuzzFeed Article"}, {"id": 790057056, "item_id": 49378566, "url": "https://news.ycombinator.com/item?id=49378566", "by": "ARandomerDude", "date": "2026-08-20", "stance": "recommend", "quote": "a coworker told me about Caveman and it has made my Claude experience vastly better.", "excerpt": "Along similar lines, a coworker told me about Caveman and it has made my Claude experience vastly better. I'm already a programmer, Claude shouldn't need to tell me like I'm 5. https://github.com/JuliusBrussee/caveman", "context": "programmer", "use_case": "terse Claude responses", "switched_from": "", "story": "Clean up Claude 5's token vomit with a separate LLM"}, {"id": 788890817, "item_id": 49305676, "url": "https://news.ycombinator.com/item?id=49305676", "by": "sjanes", "date": "2026-08-14", "stance": "neutral", "quote": "plus maybe \"caveman\" mode to cull the verbosity", "excerpt": "/handoff is the skill I use when I plainly are about to run out of token budget to give to a different harness system. Using this to avoid long sessions is something I will need to try, plus maybe \"caveman\" mode to cull the verbosity.", "context": "", "use_case": "reducing output verbosity", "switched_from": "", "story": "Maximizing the value of your Claude Code sessions"}, {"id": 788844177, "item_id": 49302761, "url": "https://news.ycombinator.com/item?id=49302761", "by": "mrinterweb", "date": "2026-08-14", "stance": "recommend", "quote": "I use the caveman skill, and I think that does help some.", "excerpt": "The pain points in the article do not bother me. I'm bothered by Opus 5's verbosity. It is so long-winded and you have to read through verbose outputs to mentally distill what is important. It is exhausting. I don't think I've ever started skim reading LLM output more that I do with Opus 5. I use the caveman skill, and I think that does help some.", "context": "", "use_case": "reducing verbosity", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}, {"id": 788283776, "item_id": 49267736, "url": "https://news.ycombinator.com/item?id=49267736", "by": "ApolloRising", "date": "2026-08-11", "stance": "recommend", "quote": "This is what you want: https://github.com/juliusbrussee/caveman", "excerpt": "This is what you want: https://github.com/juliusbrussee/caveman", "context": "", "use_case": "reduce Opus verbosity in Claude Code", "switched_from": "", "story": "Ask HN: Anyone have solution to Opus verbosity in Claude Code?"}, {"id": 784802880, "item_id": 49050180, "url": "https://news.ycombinator.com/item?id=49050180", "by": "stefanfisk", "date": "2026-07-25", "stance": "recommend", "quote": "But I still lean towards https://github.com/juliusbrussee/caveman.", "excerpt": "Neat! But I still lean towards https://github.com/juliusbrussee/caveman.", "context": "", "use_case": "making AI stop acting human", "switched_from": "", "story": "A system prompt to get AI to stop pretending to be human"}, {"id": 782488896, "item_id": 48905556, "url": "https://news.ycombinator.com/item?id=48905556", "by": "bunderbunder", "date": "2026-07-14", "stance": "recommend", "quote": "I recently started using caveman, and it\u2019s been great.", "excerpt": "I recently started using caveman, and it\u2019s been great. It doesn\u2019t just cut down on overuse of specific terms; it cuts down on time spent digesting slop in general. https://github.com/JuliusBrussee/caveman", "context": "", "use_case": "cutting verbose AI output", "switched_from": "", "story": "How to stop Claude from saying load-bearing"}, {"id": 781440880, "item_id": 48840055, "url": "https://news.ycombinator.com/item?id=48840055", "by": "hoppp", "date": "2026-07-08", "stance": "recommend", "quote": "It's for getting it to output shorter answers, but also could help with your burnout.", "excerpt": "Tell the llm to answer like a cavemen, if llm talk like cavemen, the answers become shorter and more compressed. https://github.com/JuliusBrussee/caveman It's for getting it to output shorter answers, but also could help with your burnout.", "context": "", "use_case": "shorter LLM answers", "switched_from": "", "story": "I Think I Have LLM Burnout"}, {"id": 767897600, "item_id": 47993600, "url": "https://news.ycombinator.com/item?id=47993600", "by": "anderber", "date": "2026-05-02", "stance": "recommend", "quote": "I use caveman to help with that", "excerpt": "Absolutely. I use caveman to help with that: https://github.com/JuliusBrussee/caveman", "context": "", "use_case": "reducing verbose LLM output", "switched_from": "", "story": "Kimi K2.6 just beat Claude, GPT-5.5, and Gemini in a coding challenge"}]}