{"slug": "srt", "name": "SRT", "category": "ai", "category_name": "AI models and assistants", "description": "", "recommenders": 3, "commenters": 5, "warners": 1, "score": 2.132, "rising": 0.35, "first": "2026-02-17", "latest": "2026-08-05", "why": "I launch with `srt pi` and get file system and network isolation.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 5, "neutral": 1, "compare": 0, "warn": 1}, "spark": [0, 0, 0, 0, 1, 1, 1, 1, 0, 1, 0, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "I launch with `srt pi` and get file system and network isolation.", "kind": "", "icon": false, "domain": null, "github_repo": "github.com/space-bacon/srt", "official_url": "https://github.com/space-bacon/srt", "pricing": "", "aliases": ["SRT", "srt"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 786888929, "item_id": 49180558, "url": "https://news.ycombinator.com/item?id=49180558", "by": "thehours", "date": "2026-08-05", "stance": "recommend", "quote": "I launch with `srt pi` and get file system and network isolation.", "excerpt": "I've been satisfied with using Pi inside of srt [0]. I launch with `srt pi` and get file system and network isolation. There is a seemingly infinite risk surface area to protect, but I think does a reasonable job of balancing security and convenience. I've also heard good things about nono [1] from colleagues, but I haven't personally tried it out yet. [0] https://github.com/anthropic-experimental/sandbox-runtime [1]", "context": "", "use_case": "sandboxing agents", "switched_from": "", "story": "Pi's Minimalism Is Its Advantage"}, {"id": 775500128, "item_id": 48468758, "url": "https://news.ycombinator.com/item?id=48468758", "by": "spacebacon", "date": "2026-06-09", "stance": "recommend", "quote": "SRT explains exactly how any model arrives to any answer.", "excerpt": "SRT explains exactly how any model arrives to any answer. Holding knowledge hostage is futile. https://github.com/space-bacon/SRT", "context": "", "use_case": "explaining model answers", "switched_from": "", "story": "Claude Fable 5 will sabotage \"frontier LLM research\" tasks"}, {"id": 774923568, "item_id": 48432723, "url": "https://news.ycombinator.com/item?id=48432723", "by": "spacebacon", "date": "2026-06-07", "stance": "recommend", "quote": "Leveraging a better way. No last mile.", "excerpt": "Leveraging a better way. No last mile. https://github.com/space-bacon/SRT", "context": "", "use_case": "", "switched_from": "", "story": "Harness engineering: Leveraging Codex in an agent-first world"}, {"id": 774768944, "item_id": 48423059, "url": "https://news.ycombinator.com/item?id=48423059", "by": "spacebacon", "date": "2026-06-06", "stance": "recommend", "quote": "One pass with the SRT wipes their moat out overnight.", "excerpt": "Smart choice imo. One pass with the SRT wipes their moat out overnight. https://github.com/space-bacon/SRT", "context": "", "use_case": "", "switched_from": "", "story": "S&P 500 rejects SpaceX, also blocking entry for OpenAI and Anthropic"}, {"id": 774755584, "item_id": 48422224, "url": "https://news.ycombinator.com/item?id=48422224", "by": "spacebacon", "date": "2026-06-06", "stance": "recommend", "quote": "This is the only repo that can tell you that.", "excerpt": "But how do they \u201cthink\u201d? This is the only repo that can tell you that. https://github.com/space-bacon/SRT", "context": "", "use_case": "inspecting how LLMs think", "switched_from": "", "story": "How LLMs Work"}]}], "caveats": [{"label": "", "summary": "", "quotes": [{"id": 752723554, "item_id": 47045222, "url": "https://news.ycombinator.com/item?id=47045222", "by": "alexhans", "date": "2026-02-17", "stance": "warn", "quote": "bwrap / srt still feel somewhat cumbersome", "excerpt": "This is great. I really want to find simple secure defaults when I share people how to eval [1] and bwrap / srt still feel somewhat cumbersome if you think about non tech roles. Do you have any information on estimated overhead? Information on the tradeoff of max parallelism and security options in a given system doing this vs bwrap? - [1] https://github.com/Alexhans/eval-ception", "context": "", "use_case": "", "switched_from": "", "story": "Running NanoClaw in a Docker Shell Sandbox"}]}], "who": [], "tags": [], "use_cases": ["llm adapter without fine-tuning", "sandboxing agents", "explaining model answers", "inspecting how llms think", "reading model internal thoughts", "improving recall"], "compared": [], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=49176038", "title": "Pi's Minimalism Is Its Advantage", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48467865", "title": "Claude Fable 5 will sabotage \"frontier LLM research\" tasks", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48416264", "title": "Harness engineering: Leveraging Codex in an agent-first world", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48421442", "title": "S&P 500 rejects SpaceX, also blocking entry for OpenAI and Anthropic", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48389360", "title": "How LLMs Work", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48415428", "title": "Anthropic proposes a global slowdown of AI development", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {"warn": 1, "neutral": 1}, "2026-03": {"recommend": 1}, "2026-04": {"recommend": 1}, "2026-05": {"recommend": 1}, "2026-06": {"recommend": 1}, "2026-07": {}, "2026-08": {"recommend": 1}, "2026-09": {}, "2026-10": {}}, "quotes": [{"id": 786888929, "item_id": 49180558, "url": "https://news.ycombinator.com/item?id=49180558", "by": "thehours", "date": "2026-08-05", "stance": "recommend", "quote": "I launch with `srt pi` and get file system and network isolation.", "excerpt": "I've been satisfied with using Pi inside of srt [0]. I launch with `srt pi` and get file system and network isolation. There is a seemingly infinite risk surface area to protect, but I think does a reasonable job of balancing security and convenience. I've also heard good things about nono [1] from colleagues, but I haven't personally tried it out yet. [0] https://github.com/anthropic-experimental/sandbox-runtime [1]", "context": "", "use_case": "sandboxing agents", "switched_from": "", "story": "Pi's Minimalism Is Its Advantage"}, {"id": 775500128, "item_id": 48468758, "url": "https://news.ycombinator.com/item?id=48468758", "by": "spacebacon", "date": "2026-06-09", "stance": "recommend", "quote": "SRT explains exactly how any model arrives to any answer.", "excerpt": "SRT explains exactly how any model arrives to any answer. Holding knowledge hostage is futile. https://github.com/space-bacon/SRT", "context": "", "use_case": "explaining model answers", "switched_from": "", "story": "Claude Fable 5 will sabotage \"frontier LLM research\" tasks"}, {"id": 774923568, "item_id": 48432723, "url": "https://news.ycombinator.com/item?id=48432723", "by": "spacebacon", "date": "2026-06-07", "stance": "recommend", "quote": "Leveraging a better way. No last mile.", "excerpt": "Leveraging a better way. No last mile. https://github.com/space-bacon/SRT", "context": "", "use_case": "", "switched_from": "", "story": "Harness engineering: Leveraging Codex in an agent-first world"}, {"id": 774768944, "item_id": 48423059, "url": "https://news.ycombinator.com/item?id=48423059", "by": "spacebacon", "date": "2026-06-06", "stance": "recommend", "quote": "One pass with the SRT wipes their moat out overnight.", "excerpt": "Smart choice imo. One pass with the SRT wipes their moat out overnight. https://github.com/space-bacon/SRT", "context": "", "use_case": "", "switched_from": "", "story": "S&P 500 rejects SpaceX, also blocking entry for OpenAI and Anthropic"}, {"id": 774755584, "item_id": 48422224, "url": "https://news.ycombinator.com/item?id=48422224", "by": "spacebacon", "date": "2026-06-06", "stance": "recommend", "quote": "This is the only repo that can tell you that.", "excerpt": "But how do they \u201cthink\u201d? This is the only repo that can tell you that. https://github.com/space-bacon/SRT", "context": "", "use_case": "inspecting how LLMs think", "switched_from": "", "story": "How LLMs Work"}, {"id": 774650288, "item_id": 48415643, "url": "https://news.ycombinator.com/item?id=48415643", "by": "spacebacon", "date": "2026-06-05", "stance": "recommend", "quote": "SRT can read their models thoughts internally at every step if they decide to use it.", "excerpt": "SRT can read their models thoughts internally at every step if they decide to use it. The problem is that it lowers a draw bridge over their proprietary moat. So of course it\u2019s a natural time to signal a global slowdown. https://github.com/space-bacon/SRT", "context": "", "use_case": "reading model internal thoughts", "switched_from": "", "story": "Anthropic proposes a global slowdown of AI development"}, {"id": 774551776, "item_id": 48409486, "url": "https://news.ycombinator.com/item?id=48409486", "by": "spacebacon", "date": "2026-06-05", "stance": "recommend", "quote": "Now do it without the fine tuning.", "excerpt": "Now do it without the fine tuning. https://github.com/space-bacon/SRT The HF zool4nd3r demo may be useful", "context": "", "use_case": "LLM adapter without fine-tuning", "switched_from": "", "story": "Fine-tuning an LLM to write docs like it's 1995"}, {"id": 774551040, "item_id": 48409440, "url": "https://news.ycombinator.com/item?id=48409440", "by": "spacebacon", "date": "2026-06-05", "stance": "recommend", "quote": "No fine tuning required. You can use the SRT adapter.", "excerpt": "No fine tuning required. You can use the SRT adapter. https://github.com/space-bacon/SRT", "context": "", "use_case": "LLM adapter without fine-tuning", "switched_from": "", "story": "Latent Agents: A Post-Training Procedure for Internalized Multi-Agent Debate"}, {"id": 774469600, "item_id": 48404350, "url": "https://news.ycombinator.com/item?id=48404350", "by": "spacebacon", "date": "2026-06-04", "stance": "recommend", "quote": "This one is and should be adapted to every frozen model ASAP.", "excerpt": "This one is and should be adapted to every frozen model ASAP. https://github.com/space-bacon/SRT Significantly improve every frozen model overnight. LFG.", "context": "", "use_case": "", "switched_from": "", "story": "Anthropic's open-source framework for AI-powered vulnerability discovery"}, {"id": 769084592, "item_id": 48067787, "url": "https://news.ycombinator.com/item?id=48067787", "by": "spacebacon", "date": "2026-05-08", "stance": "recommend", "quote": "This repository empirically proves computational semiotics", "excerpt": "Related: https://github.com/space-bacon/SRT This repository empirically proves computational semiotics. The only \u201cmetacognitive\u201d (2nd order) and metapragmatic (3rd order) model I\u2019m aware of.", "context": "", "use_case": "", "switched_from": "", "story": "Hallucinations Undermine Trust; Metacognition Is a Way Forward"}, {"id": 767453216, "item_id": 47965826, "url": "https://news.ycombinator.com/item?id=47965826", "by": "spacebacon", "date": "2026-04-30", "stance": "recommend", "quote": "For recall, may want to check out the SRT.", "excerpt": "For recall, may want to check out the SRT. https://huggingface.co/spaces/RiverRider/srt-adapter-v8a-demo", "context": "", "use_case": "improving recall", "switched_from": "", "story": "Finetuning Activates Verbatim Recall of Copyrighted Books in LLMs"}, {"id": 759502193, "item_id": 47468887, "url": "https://news.ycombinator.com/item?id=47468887", "by": "kmod", "date": "2026-03-21", "stance": "recommend", "quote": "I use srt and it was pretty straightforward to set up", "excerpt": "Fwiw this got changed about a week ago, where they changed the logic to match the documentation rather than default to sending your prompts to their servers. This is why so many people have noticed this happening but if you ask an AI about it right now it will say this is not true. Personally I think it's necessary to run opencode itself inside a sandbox, and if you do that you can see all of the rejected network cal", "context": "", "use_case": "Sandboxing agent processes", "switched_from": "", "story": "OpenCode \u2013 The open source AI coding agent"}, {"id": 753495632, "item_id": 47093477, "url": "https://news.ycombinator.com/item?id=47093477", "by": "zsiec", "date": "2026-02-20", "stance": "neutral", "quote": "Ported from https://github.com/Haivision/srt with interop and bench tests", "excerpt": "Ported from https://github.com/Haivision/srt with interop and bench tests", "context": "", "use_case": "", "switched_from": "", "story": "No Compromise Pure Golang Version of Haivision's SRT (Secure Reliable Transport)"}, {"id": 752723554, "item_id": 47045222, "url": "https://news.ycombinator.com/item?id=47045222", "by": "alexhans", "date": "2026-02-17", "stance": "warn", "quote": "bwrap / srt still feel somewhat cumbersome", "excerpt": "This is great. I really want to find simple secure defaults when I share people how to eval [1] and bwrap / srt still feel somewhat cumbersome if you think about non tech roles. Do you have any information on estimated overhead? Information on the tradeoff of max parallelism and security options in a given system doing this vs bwrap? - [1] https://github.com/Alexhans/eval-ception", "context": "", "use_case": "", "switched_from": "", "story": "Running NanoClaw in a Docker Shell Sandbox"}]}