{"slug": "deepseek-harness", "name": "DeepSeek Harness", "category": "ai-coding", "category_name": "AI coding tools", "description": "An open-source AI coding agent harness from DeepSeek, with a web UI and session tracing.", "recommenders": 6, "commenters": 8, "warners": 0, "score": 6.474, "rising": 6.0, "first": "2026-08-13", "latest": "2026-10-26", "why": "Liked for fast, low-CPU runs and a well-thought-out architecture, with cheap results on generative tasks and support for providers like OpenRouter.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 7, "neutral": 2, "compare": 0, "warn": 0}, "spark": [0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 4], "listed": true, "loved": ["Speed", "Architecture"], "caveat": "Token usage", "short_quote": "I am using DeepSeek Harness[0] (switched from OpenCode)", "icon": true, "domain": "deepseek.com", "github_repo": null, "official_url": "https://deepseek.com/harness/en/", "pricing": "open source", "aliases": ["DeepSeek Harness", "DeepSeek harness", "Deepseek Harness", "deepseek-harness"], "verdict": [{"text": "Liked for fast, low-CPU runs and a well-thought-out architecture", "cites": [{"id": 802082992, "item_id": 50130187, "url": "https://news.ycombinator.com/item?id=50130187", "by": "ggeorgovassilis", "date": "2026-10-17", "stance": "recommend", "quote": "It's fast, solid, I love the architecture", "excerpt": "I started using it last week when copilot started acting up against the Deepseek models. It's fast, solid, I love the architecture but be warned: it submits very very many input tokens against the LLM endpoint, so robust & cheap token caching is a must.", "context": "", "use_case": "agent harness for DeepSeek models", "switched_from": "Copilot", "story": "Ultramarine 1\u00a0Flash"}, {"id": 796929792, "item_id": 49808112, "url": "https://news.ycombinator.com/item?id=49808112", "by": "wg0", "date": "2026-09-22", "stance": "recommend", "quote": "I am using DeepSeek Harness[0] (switched from OpenCode)", "excerpt": "I am using DeepSeek Harness[0] (switched from OpenCode) and I am using DeepSeek directly via the API. The speed is insane. Like 200 tokens/second is the norm but I have seen much higher too at times. PS: I do not know why but opencode pushes CPU usage to very high which has NOT happened with DeepSeek harness even once. [0]. https://github.com/deepseek-ai/deepseek-harness", "context": "", "use_case": "coding agent harness", "switched_from": "OpenCode", "story": "Claude Opus 5.5"}, {"id": 798964176, "item_id": 49935261, "url": "https://news.ycombinator.com/item?id=49935261", "by": "wren6991", "date": "2026-10-02", "stance": "recommend", "quote": "I think DSH is a pretty well-built vanilla harness with a nice web UI", "excerpt": "Seems like a category error: you can use DSH with openrouter, in fact it ships built-in support using the Pi LLM SDK. I think DSH is a pretty well-built vanilla harness with a nice web UI (I use it in a pinned browser tab). I much prefer this to dealing with TUI clipboard/scroll jank, and it works just as well over SSH if you simply forward the port. They've genuinely thought about the architecture, and made some eff", "context": "uses it in a pinned browser tab", "use_case": "LLM coding harness", "switched_from": "", "story": "DeepSeek Harness Desktop for macOS and Windows"}]}, {"text": ", with cheap results on generative tasks and support for providers like OpenRouter.", "cites": [{"id": 798207440, "item_id": 49887965, "url": "https://news.ycombinator.com/item?id=49887965", "by": "Kuyawa", "date": "2026-09-29", "stance": "recommend", "quote": "using DeepSeek Harness, spent three hours in total, $3 in tokens, and got a playable and enjoyable game", "excerpt": "I ported Prince of Persia to Swift for MacOS [1] over the weekend using DeepSeek Harness, spent three hours in total, $3 in tokens, and got a playable and enjoyable game bringing back good ol' memories It is a faithful port of Jordan Mechner's 1989 original game and the guiding idea was just one sentence: port the game exactly, rewrite the engine completely I didn't write a single line of code or test, not even the r", "context": "", "use_case": "porting a game to Swift", "switched_from": "", "story": "What reversing, modernising old games tells us about the economic impact of AI"}, {"id": 804035378, "item_id": 50252211, "url": "https://news.ycombinator.com/item?id=50252211", "by": "anigbrowl", "date": "2026-10-26", "stance": "recommend", "quote": "eventually I switched to Deepseek Harness for generative tasks", "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": "generative coding tasks", "switched_from": "Cline", "story": "JetBrains Thinkrail"}, {"id": 798964176, "item_id": 49935261, "url": "https://news.ycombinator.com/item?id=49935261", "by": "wren6991", "date": "2026-10-02", "stance": "recommend", "quote": "I think DSH is a pretty well-built vanilla harness with a nice web UI", "excerpt": "Seems like a category error: you can use DSH with openrouter, in fact it ships built-in support using the Pi LLM SDK. I think DSH is a pretty well-built vanilla harness with a nice web UI (I use it in a pinned browser tab). I much prefer this to dealing with TUI clipboard/scroll jank, and it works just as well over SSH if you simply forward the port. They've genuinely thought about the architecture, and made some eff", "context": "uses it in a pinned browser tab", "use_case": "LLM coding harness", "switched_from": "", "story": "DeepSeek Harness Desktop for macOS and Windows"}, {"id": 788744896, "item_id": 49296556, "url": "https://news.ycombinator.com/item?id=49296556", "by": "brhenc", "date": "2026-08-14", "stance": "neutral", "quote": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost.", "excerpt": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost. I would be happier if it wasn't javascript-based, but it's open source so it should be doable to use the models with things like https://github.com/charmbracelet/crush and optimize a little how these alternatives work with deepseek models. For prototyping the web interface is OK, but I don't", "context": "lives in the terminal", "use_case": "coding harness", "switched_from": "", "story": "DeepSeek Harness official website launched"}]}, {"text": " Caveats include very heavy input-token usage that makes token caching a must", "cites": [{"id": 802082992, "item_id": 50130187, "url": "https://news.ycombinator.com/item?id=50130187", "by": "ggeorgovassilis", "date": "2026-10-17", "stance": "recommend", "quote": "It's fast, solid, I love the architecture", "excerpt": "I started using it last week when copilot started acting up against the Deepseek models. It's fast, solid, I love the architecture but be warned: it submits very very many input tokens against the LLM endpoint, so robust & cheap token caching is a must.", "context": "", "use_case": "agent harness for DeepSeek models", "switched_from": "Copilot", "story": "Ultramarine 1\u00a0Flash"}]}, {"text": ", and a JavaScript-based web interface that terminal users find less pleasant.", "cites": [{"id": 788744896, "item_id": 49296556, "url": "https://news.ycombinator.com/item?id=49296556", "by": "brhenc", "date": "2026-08-14", "stance": "neutral", "quote": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost.", "excerpt": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost. I would be happier if it wasn't javascript-based, but it's open source so it should be doable to use the models with things like https://github.com/charmbracelet/crush and optimize a little how these alternatives work with deepseek models. For prototyping the web interface is OK, but I don't", "context": "lives in the terminal", "use_case": "coding harness", "switched_from": "", "story": "DeepSeek Harness official website launched"}]}], "praise": [{"label": "Speed", "summary": "Commenters call it fast, report around 200 tokens/second via the DeepSeek API, and note low CPU use compared to OpenCode.", "quotes": [{"id": 796929792, "item_id": 49808112, "url": "https://news.ycombinator.com/item?id=49808112", "by": "wg0", "date": "2026-09-22", "stance": "recommend", "quote": "I am using DeepSeek Harness[0] (switched from OpenCode)", "excerpt": "I am using DeepSeek Harness[0] (switched from OpenCode) and I am using DeepSeek directly via the API. The speed is insane. Like 200 tokens/second is the norm but I have seen much higher too at times. PS: I do not know why but opencode pushes CPU usage to very high which has NOT happened with DeepSeek harness even once. [0]. https://github.com/deepseek-ai/deepseek-harness", "context": "", "use_case": "coding agent harness", "switched_from": "OpenCode", "story": "Claude Opus 5.5"}, {"id": 802082992, "item_id": 50130187, "url": "https://news.ycombinator.com/item?id=50130187", "by": "ggeorgovassilis", "date": "2026-10-17", "stance": "recommend", "quote": "It's fast, solid, I love the architecture", "excerpt": "I started using it last week when copilot started acting up against the Deepseek models. It's fast, solid, I love the architecture but be warned: it submits very very many input tokens against the LLM endpoint, so robust & cheap token caching is a must.", "context": "", "use_case": "agent harness for DeepSeek models", "switched_from": "Copilot", "story": "Ultramarine 1\u00a0Flash"}, {"id": 801639538, "item_id": 50102471, "url": "https://news.ycombinator.com/item?id=50102471", "by": "wg0", "date": "2026-10-15", "stance": "recommend", "quote": "That's why I first switched to Pi and then to DeepSeek harness.", "excerpt": "Performance is horrible both of the website and of the tool itself. That's why I first switched to Pi and then to DeepSeek harness.", "context": "", "use_case": "AI coding agent", "switched_from": "Pi", "story": "OpenCode 2.0"}]}, {"label": "Architecture", "summary": "Described as a well-built, thoughtfully architected harness with novel concepts and some shell sandboxing.", "quotes": [{"id": 798964176, "item_id": 49935261, "url": "https://news.ycombinator.com/item?id=49935261", "by": "wren6991", "date": "2026-10-02", "stance": "recommend", "quote": "I think DSH is a pretty well-built vanilla harness with a nice web UI", "excerpt": "Seems like a category error: you can use DSH with openrouter, in fact it ships built-in support using the Pi LLM SDK. I think DSH is a pretty well-built vanilla harness with a nice web UI (I use it in a pinned browser tab). I much prefer this to dealing with TUI clipboard/scroll jank, and it works just as well over SSH if you simply forward the port. They've genuinely thought about the architecture, and made some eff", "context": "uses it in a pinned browser tab", "use_case": "LLM coding harness", "switched_from": "", "story": "DeepSeek Harness Desktop for macOS and Windows"}, {"id": 802082992, "item_id": 50130187, "url": "https://news.ycombinator.com/item?id=50130187", "by": "ggeorgovassilis", "date": "2026-10-17", "stance": "recommend", "quote": "It's fast, solid, I love the architecture", "excerpt": "I started using it last week when copilot started acting up against the Deepseek models. It's fast, solid, I love the architecture but be warned: it submits very very many input tokens against the LLM endpoint, so robust & cheap token caching is a must.", "context": "", "use_case": "agent harness for DeepSeek models", "switched_from": "Copilot", "story": "Ultramarine 1\u00a0Flash"}, {"id": 788579440, "item_id": 49286215, "url": "https://news.ycombinator.com/item?id=49286215", "by": "kamranjon", "date": "2026-08-13", "stance": "recommend", "quote": "New coding harness that seems to have some novel concepts and one of the pretty cool things", "excerpt": "New coding harness that seems to have some novel concepts and one of the pretty cool things on their landing page for it here: https://deepseek.com/harness/en/ is the Every Run is Traceable view: \"Everything the model sees is recorded in an append-only session log: system prompts, reasoning, tool calls and results, subagent scheduling, and every context injection. In the Trajectory view, you can inspect these records", "context": "", "use_case": "coding agent harness", "switched_from": "", "story": "DeepSeek Harness"}]}, {"label": "Cost efficiency", "summary": "Users report low token costs, such as a full game port for $3, and less verbose output.", "quotes": [{"id": 798207440, "item_id": 49887965, "url": "https://news.ycombinator.com/item?id=49887965", "by": "Kuyawa", "date": "2026-09-29", "stance": "recommend", "quote": "using DeepSeek Harness, spent three hours in total, $3 in tokens, and got a playable and enjoyable game", "excerpt": "I ported Prince of Persia to Swift for MacOS [1] over the weekend using DeepSeek Harness, spent three hours in total, $3 in tokens, and got a playable and enjoyable game bringing back good ol' memories It is a faithful port of Jordan Mechner's 1989 original game and the guiding idea was just one sentence: port the game exactly, rewrite the engine completely I didn't write a single line of code or test, not even the r", "context": "", "use_case": "porting a game to Swift", "switched_from": "", "story": "What reversing, modernising old games tells us about the economic impact of AI"}, {"id": 802074018, "item_id": 50129626, "url": "https://news.ycombinator.com/item?id=50129626", "by": "nickludlam", "date": "2026-10-17", "stance": "neutral", "quote": "I use each of them in their own respective coding harnesses, deepseek-harness and kimi-code", "excerpt": "This and (the not quite yet open) Kimi K2.8 are my main coding engines, and I find them incredibly cost efficient. I find Deepseek in particular is really good with Svelte work. Right now I use each of them in their own respective coding harnesses, deepseek-harness and kimi-code, but I do wonder what they are like inside something like opencode.", "context": "", "use_case": "running DeepSeek for coding", "switched_from": "", "story": "DeepSeek-v4.1-Flash-1017"}, {"id": 788744896, "item_id": 49296556, "url": "https://news.ycombinator.com/item?id=49296556", "by": "brhenc", "date": "2026-08-14", "stance": "neutral", "quote": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost.", "excerpt": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost. I would be happier if it wasn't javascript-based, but it's open source so it should be doable to use the models with things like https://github.com/charmbracelet/crush and optimize a little how these alternatives work with deepseek models. For prototyping the web interface is OK, but I don't", "context": "lives in the terminal", "use_case": "coding harness", "switched_from": "", "story": "DeepSeek Harness official website launched"}]}, {"label": "Generative tasks", "summary": "Users have switched to it for generative coding work and report strong results from short prompts.", "quotes": [{"id": 804035378, "item_id": 50252211, "url": "https://news.ycombinator.com/item?id=50252211", "by": "anigbrowl", "date": "2026-10-26", "stance": "recommend", "quote": "eventually I switched to Deepseek Harness for generative tasks", "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": "generative coding tasks", "switched_from": "Cline", "story": "JetBrains Thinkrail"}, {"id": 798207440, "item_id": 49887965, "url": "https://news.ycombinator.com/item?id=49887965", "by": "Kuyawa", "date": "2026-09-29", "stance": "recommend", "quote": "using DeepSeek Harness, spent three hours in total, $3 in tokens, and got a playable and enjoyable game", "excerpt": "I ported Prince of Persia to Swift for MacOS [1] over the weekend using DeepSeek Harness, spent three hours in total, $3 in tokens, and got a playable and enjoyable game bringing back good ol' memories It is a faithful port of Jordan Mechner's 1989 original game and the guiding idea was just one sentence: port the game exactly, rewrite the engine completely I didn't write a single line of code or test, not even the r", "context": "", "use_case": "porting a game to Swift", "switched_from": "", "story": "What reversing, modernising old games tells us about the economic impact of AI"}]}, {"label": "Provider flexibility", "summary": "It works with OpenRouter and is not locked to DeepSeek's own API.", "quotes": [{"id": 798964176, "item_id": 49935261, "url": "https://news.ycombinator.com/item?id=49935261", "by": "wren6991", "date": "2026-10-02", "stance": "recommend", "quote": "I think DSH is a pretty well-built vanilla harness with a nice web UI", "excerpt": "Seems like a category error: you can use DSH with openrouter, in fact it ships built-in support using the Pi LLM SDK. I think DSH is a pretty well-built vanilla harness with a nice web UI (I use it in a pinned browser tab). I much prefer this to dealing with TUI clipboard/scroll jank, and it works just as well over SSH if you simply forward the port. They've genuinely thought about the architecture, and made some eff", "context": "uses it in a pinned browser tab", "use_case": "LLM coding harness", "switched_from": "", "story": "DeepSeek Harness Desktop for macOS and Windows"}, {"id": 788744896, "item_id": 49296556, "url": "https://news.ycombinator.com/item?id=49296556", "by": "brhenc", "date": "2026-08-14", "stance": "neutral", "quote": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost.", "excerpt": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost. I would be happier if it wasn't javascript-based, but it's open source so it should be doable to use the models with things like https://github.com/charmbracelet/crush and optimize a little how these alternatives work with deepseek models. For prototyping the web interface is OK, but I don't", "context": "lives in the terminal", "use_case": "coding harness", "switched_from": "", "story": "DeepSeek Harness official website launched"}]}, {"label": "Traceability", "summary": "An append-only session log supports inspecting, resuming, forking and replaying runs.", "quotes": [{"id": 788579440, "item_id": 49286215, "url": "https://news.ycombinator.com/item?id=49286215", "by": "kamranjon", "date": "2026-08-13", "stance": "recommend", "quote": "New coding harness that seems to have some novel concepts and one of the pretty cool things", "excerpt": "New coding harness that seems to have some novel concepts and one of the pretty cool things on their landing page for it here: https://deepseek.com/harness/en/ is the Every Run is Traceable view: \"Everything the model sees is recorded in an append-only session log: system prompts, reasoning, tool calls and results, subagent scheduling, and every context injection. In the Trajectory view, you can inspect these records", "context": "", "use_case": "coding agent harness", "switched_from": "", "story": "DeepSeek Harness"}]}], "caveats": [{"label": "Token usage", "summary": "It sends very many input tokens to the LLM endpoint, so robust and cheap caching is needed.", "quotes": [{"id": 802082992, "item_id": 50130187, "url": "https://news.ycombinator.com/item?id=50130187", "by": "ggeorgovassilis", "date": "2026-10-17", "stance": "recommend", "quote": "It's fast, solid, I love the architecture", "excerpt": "I started using it last week when copilot started acting up against the Deepseek models. It's fast, solid, I love the architecture but be warned: it submits very very many input tokens against the LLM endpoint, so robust & cheap token caching is a must.", "context": "", "use_case": "agent harness for DeepSeek models", "switched_from": "Copilot", "story": "Ultramarine 1\u00a0Flash"}]}, {"label": "JavaScript-based", "summary": "One commenter would prefer it not be written in JavaScript outside sandboxed environments.", "quotes": [{"id": 788744896, "item_id": 49296556, "url": "https://news.ycombinator.com/item?id=49296556", "by": "brhenc", "date": "2026-08-14", "stance": "neutral", "quote": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost.", "excerpt": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost. I would be happier if it wasn't javascript-based, but it's open source so it should be doable to use the models with things like https://github.com/charmbracelet/crush and optimize a little how these alternatives work with deepseek models. For prototyping the web interface is OK, but I don't", "context": "lives in the terminal", "use_case": "coding harness", "switched_from": "", "story": "DeepSeek Harness official website launched"}]}, {"label": "Web UI", "summary": "A terminal-focused user found the web interface less nice to use, though OK for prototyping.", "quotes": [{"id": 788744896, "item_id": 49296556, "url": "https://news.ycombinator.com/item?id=49296556", "by": "brhenc", "date": "2026-08-14", "stance": "neutral", "quote": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost.", "excerpt": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost. I would be happier if it wasn't javascript-based, but it's open source so it should be doable to use the models with things like https://github.com/charmbracelet/crush and optimize a little how these alternatives work with deepseek models. For prototyping the web interface is OK, but I don't", "context": "lives in the terminal", "use_case": "coding harness", "switched_from": "", "story": "DeepSeek Harness official website launched"}]}], "who": [{"by": "wren6991", "context": "uses it in a pinned browser tab", "url": "https://news.ycombinator.com/item?id=49935261", "date": "2026-10-02"}, {"by": "brhenc", "context": "lives in the terminal", "url": "https://news.ycombinator.com/item?id=49296556", "date": "2026-08-14"}], "tags": ["open source", "works with OpenRouter", "for DeepSeek API users", "usable over SSH"], "use_cases": ["coding agent harness", "generative coding tasks", "agent harness for deepseek models", "running deepseek for coding", "ai coding agent", "llm coding harness"], "compared": [{"slug": "opencode", "name": "OpenCode", "co": 3, "switched_from": 1, "switched_to": 0}, {"slug": "pi", "name": "Pi", "co": 2, "switched_from": 1, "switched_to": 0}, {"slug": "copilot", "name": "Copilot", "co": 1, "switched_from": 1, "switched_to": 0}, {"slug": "cline", "name": "Cline", "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=50106747", "title": "Ultramarine 1\u00a0Flash", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=50129045", "title": "DeepSeek-v4.1-Flash-1017", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=50094994", "title": "OpenCode 2.0", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49929489", "title": "DeepSeek Harness Desktop for macOS and Windows", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49861755", "title": "What reversing, modernising old games tells us about the economic impact of AI", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {}, "2026-03": {}, "2026-04": {}, "2026-05": {}, "2026-06": {}, "2026-07": {}, "2026-08": {"recommend": 1, "neutral": 1}, "2026-09": {"recommend": 2}, "2026-10": {"recommend": 4, "neutral": 1}}, "quotes": [{"id": 804035378, "item_id": 50252211, "url": "https://news.ycombinator.com/item?id=50252211", "by": "anigbrowl", "date": "2026-10-26", "stance": "recommend", "quote": "eventually I switched to Deepseek Harness for generative tasks", "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": "generative coding tasks", "switched_from": "Cline", "story": "JetBrains Thinkrail"}, {"id": 802082992, "item_id": 50130187, "url": "https://news.ycombinator.com/item?id=50130187", "by": "ggeorgovassilis", "date": "2026-10-17", "stance": "recommend", "quote": "It's fast, solid, I love the architecture", "excerpt": "I started using it last week when copilot started acting up against the Deepseek models. It's fast, solid, I love the architecture but be warned: it submits very very many input tokens against the LLM endpoint, so robust & cheap token caching is a must.", "context": "", "use_case": "agent harness for DeepSeek models", "switched_from": "Copilot", "story": "Ultramarine 1\u00a0Flash"}, {"id": 802074018, "item_id": 50129626, "url": "https://news.ycombinator.com/item?id=50129626", "by": "nickludlam", "date": "2026-10-17", "stance": "neutral", "quote": "I use each of them in their own respective coding harnesses, deepseek-harness and kimi-code", "excerpt": "This and (the not quite yet open) Kimi K2.8 are my main coding engines, and I find them incredibly cost efficient. I find Deepseek in particular is really good with Svelte work. Right now I use each of them in their own respective coding harnesses, deepseek-harness and kimi-code, but I do wonder what they are like inside something like opencode.", "context": "", "use_case": "running DeepSeek for coding", "switched_from": "", "story": "DeepSeek-v4.1-Flash-1017"}, {"id": 801639538, "item_id": 50102471, "url": "https://news.ycombinator.com/item?id=50102471", "by": "wg0", "date": "2026-10-15", "stance": "recommend", "quote": "That's why I first switched to Pi and then to DeepSeek harness.", "excerpt": "Performance is horrible both of the website and of the tool itself. That's why I first switched to Pi and then to DeepSeek harness.", "context": "", "use_case": "AI coding agent", "switched_from": "Pi", "story": "OpenCode 2.0"}, {"id": 798964176, "item_id": 49935261, "url": "https://news.ycombinator.com/item?id=49935261", "by": "wren6991", "date": "2026-10-02", "stance": "recommend", "quote": "I think DSH is a pretty well-built vanilla harness with a nice web UI", "excerpt": "Seems like a category error: you can use DSH with openrouter, in fact it ships built-in support using the Pi LLM SDK. I think DSH is a pretty well-built vanilla harness with a nice web UI (I use it in a pinned browser tab). I much prefer this to dealing with TUI clipboard/scroll jank, and it works just as well over SSH if you simply forward the port. They've genuinely thought about the architecture, and made some eff", "context": "uses it in a pinned browser tab", "use_case": "LLM coding harness", "switched_from": "", "story": "DeepSeek Harness Desktop for macOS and Windows"}, {"id": 798207440, "item_id": 49887965, "url": "https://news.ycombinator.com/item?id=49887965", "by": "Kuyawa", "date": "2026-09-29", "stance": "recommend", "quote": "using DeepSeek Harness, spent three hours in total, $3 in tokens, and got a playable and enjoyable game", "excerpt": "I ported Prince of Persia to Swift for MacOS [1] over the weekend using DeepSeek Harness, spent three hours in total, $3 in tokens, and got a playable and enjoyable game bringing back good ol' memories It is a faithful port of Jordan Mechner's 1989 original game and the guiding idea was just one sentence: port the game exactly, rewrite the engine completely I didn't write a single line of code or test, not even the r", "context": "", "use_case": "porting a game to Swift", "switched_from": "", "story": "What reversing, modernising old games tells us about the economic impact of AI"}, {"id": 796929792, "item_id": 49808112, "url": "https://news.ycombinator.com/item?id=49808112", "by": "wg0", "date": "2026-09-22", "stance": "recommend", "quote": "I am using DeepSeek Harness[0] (switched from OpenCode)", "excerpt": "I am using DeepSeek Harness[0] (switched from OpenCode) and I am using DeepSeek directly via the API. The speed is insane. Like 200 tokens/second is the norm but I have seen much higher too at times. PS: I do not know why but opencode pushes CPU usage to very high which has NOT happened with DeepSeek harness even once. [0]. https://github.com/deepseek-ai/deepseek-harness", "context": "", "use_case": "coding agent harness", "switched_from": "OpenCode", "story": "Claude Opus 5.5"}, {"id": 788744896, "item_id": 49296556, "url": "https://news.ycombinator.com/item?id=49296556", "by": "brhenc", "date": "2026-08-14", "stance": "neutral", "quote": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost.", "excerpt": "Tried it this morning - it works with openrouter, feels a little less verbose than antigravity CLI which should help with cost. I would be happier if it wasn't javascript-based, but it's open source so it should be doable to use the models with things like https://github.com/charmbracelet/crush and optimize a little how these alternatives work with deepseek models. For prototyping the web interface is OK, but I don't", "context": "lives in the terminal", "use_case": "coding harness", "switched_from": "", "story": "DeepSeek Harness official website launched"}, {"id": 788579440, "item_id": 49286215, "url": "https://news.ycombinator.com/item?id=49286215", "by": "kamranjon", "date": "2026-08-13", "stance": "recommend", "quote": "New coding harness that seems to have some novel concepts and one of the pretty cool things", "excerpt": "New coding harness that seems to have some novel concepts and one of the pretty cool things on their landing page for it here: https://deepseek.com/harness/en/ is the Every Run is Traceable view: \"Everything the model sees is recorded in an append-only session log: system prompts, reasoning, tool calls and results, subagent scheduling, and every context injection. In the Trajectory view, you can inspect these records", "context": "", "use_case": "coding agent harness", "switched_from": "", "story": "DeepSeek Harness"}]}