Olivia
Online
fatherrigby
@fatherrigby
trench junkie in search for internet money
Joined May 2026
598 Following
57 Followers
394 Posts
Pinned Tweet
DeepSeek just released something interesting
It’s called DeepSeek Harness (DSH), and the important part isn’t just what it can do out of the box.
It’s the architecture.
The easiest way to understand it is:
Model → Harness → Plugins
The model provides the reasoning.
The Harness gives that reasoning an environment to actually operate in: file editing, shell execution, search, planning, subagents, workflows, sessions, permissions and more.
And underneath it is Cordis, the plugin runtime that makes the system composable.
Instead of building one giant, hard-coded agent, DSH treats many of the components that make an agent useful as plugins.
That means developers can build, replace and compose capabilities instead of waiting for the core agent to implement everything.
The Harness comes with several runtime modes:
Standard Mode: the full coding-agent experience with tools, planning, skills, subagents and workflows.
Code Mode: lets the model orchestrate tool operations through generated TypeScript, making complex multi-step tool use programmable.
Minimal Mode: strips the agent down to persistent Bash + a file editor.
Creator Mode: exposes the runtime for experimenting with plugins and creating custom agent presets.
The session architecture is another interesting piece.
DSH records the agent's execution as an append-only trajectory: reasoning, tool calls, tool results, subagent activity and context changes.
Sessions can then be resumed, forked, searched and replayed.
So an agent run isn't just a chat that disappears when you're done. It's an inspectable execution history.
And the ecosystem is already being built around this.
DeepSeek has a "dsh-plugin" ecosystem where developers can create additional capabilities for the Harness.
It can also work with other providers and custom OpenAI-compatible endpoints, so the Harness isn't simply a UI wrapped around one DeepSeek model.
The bigger picture is what makes this interesting:
The model is the brain.
The Harness is the body.
Plugins are the extensibility layer.
DeepSeek isn't just giving developers another agent to use.
They're giving developers an environment for building custom agents.
It's still a developer preview and the architecture is evolving There's no runner for it yet so I'm deploying $DSH
@vectorfekh 2yuZ5qUHYKkaKZ81ZHgjMShoFx6PoSfoaioJBirjpump
Neko no Therapists! Looks like it's straight out of an anime
1m views on it too lets send $fumifumi
일본 도쿄에 있다는 고양이 마사지 샵😆
2yuZ5qUHYKkaKZ81ZHgjMShoFx6PoSfoaioJBirjpump
Ca: 2yuZ5qUHYKkaKZ81ZHgjMShoFx6PoSfoaioJBirjpump
@godfather_hq good morning chief it gets better
@TheBabylonBee And he got gyatt!!!
https://t.co/mXHg8Auj5L
Send the $GYATTIMUS
3AdRvjJrye4WwuY3vgbfupocCCjAzauFJUBcJ3N2pump
no way we got twerking optimus, low quality meme videos been sending lately this should follow soon, let's send $GYATTIMUS
@xbtkatsu https://t.co/mXHg8Auj5L
Send the $GYATTIMUS
3AdRvjJrye4WwuY3vgbfupocCCjAzauFJUBcJ3N2pump
no way we got twerking optimus, low quality meme videos been sending lately this should follow soon, let's send $GYATTIMUS
@Defipetros101 Too real bro😭
3AdRvjJrye4WwuY3vgbfupocCCjAzauFJUBcJ3N2pump
no way we got twerking optimus, low quality meme videos been sending lately this should follow soon, let's send $GYATTIMUS
Ca 3AdRvjJrye4WwuY3vgbfupocCCjAzauFJUBcJ3N2pump
DeepSeek just released something interesting
It’s called DeepSeek Harness (DSH), and the important part isn’t just what it can do out of the box.
It’s the architecture.
The easiest way to understand it is:
Model → Harness → Plugins
The model provides the reasoning.
The Harness gives that reasoning an environment to actually operate in: file editing, shell execution, search, planning, subagents, workflows, sessions, permissions and more.
And underneath it is Cordis, the plugin runtime that makes the system composable.
Instead of building one giant, hard-coded agent, DSH treats many of the components that make an agent useful as plugins.
That means developers can build, replace and compose capabilities instead of waiting for the core agent to implement everything.
The Harness comes with several runtime modes:
Standard Mode: the full coding-agent experience with tools, planning, skills, subagents and workflows.
Code Mode: lets the model orchestrate tool operations through generated TypeScript, making complex multi-step tool use programmable.
Minimal Mode: strips the agent down to persistent Bash + a file editor.
Creator Mode: exposes the runtime for experimenting with plugins and creating custom agent presets.
The session architecture is another interesting piece.
DSH records the agent's execution as an append-only trajectory: reasoning, tool calls, tool results, subagent activity and context changes.
Sessions can then be resumed, forked, searched and replayed.
So an agent run isn't just a chat that disappears when you're done. It's an inspectable execution history.
And the ecosystem is already being built around this.
DeepSeek has a "dsh-plugin" ecosystem where developers can create additional capabilities for the Harness.
It can also work with other providers and custom OpenAI-compatible endpoints, so the Harness isn't simply a UI wrapped around one DeepSeek model.
The bigger picture is what makes this interesting:
The model is the brain.
The Harness is the body.
Plugins are the extensibility layer.
DeepSeek isn't just giving developers another agent to use.
They're giving developers an environment for building custom agents.
It's still a developer preview and the architecture is evolving There's no runner for it yet so I'm deploying $DSH
DeepSeek just released Harness, multiple steps improvement in coding agents, instead of releasing one hard-coded agent $DSH treats many of the components that make an agent useful as plugins.. that means developers can build, replace and compose capabilities instead of waiting for DeepSeek to implement the features they need everything in $DSH is literally a plugin
https://t.co/za7SV0bZul
CA: EQbtWcRQ8R6uZvgvaoAEEchRJJTdm1p2nuDNCht7pump
https://t.co/DCIbryfic3
DeepSeek just released something interesting
It’s called DeepSeek Harness (DSH), and the important part isn’t just what it can do out of the box.
It’s the architecture.
The easiest way to understand it is:
Model → Harness → Plugins
The model provides the reasoning.
The Harness gives that reasoning an environment to actually operate in: file editing, shell execution, search, planning, subagents, workflows, sessions, permissions and more.
And underneath it is Cordis, the plugin runtime that makes the system composable.
Instead of building one giant, hard-coded agent, DSH treats many of the components that make an agent useful as plugins.
That means developers can build, replace and compose capabilities instead of waiting for the core agent to implement everything.
The Harness comes with several runtime modes:
Standard Mode: the full coding-agent experience with tools, planning, skills, subagents and workflows.
Code Mode: lets the model orchestrate tool operations through generated TypeScript, making complex multi-step tool use programmable.
Minimal Mode: strips the agent down to persistent Bash + a file editor.
Creator Mode: exposes the runtime for experimenting with plugins and creating custom agent presets.
The session architecture is another interesting piece.
DSH records the agent's execution as an append-only trajectory: reasoning, tool calls, tool results, subagent activity and context changes.
Sessions can then be resumed, forked, searched and replayed.
So an agent run isn't just a chat that disappears when you're done. It's an inspectable execution history.
And the ecosystem is already being built around this.
DeepSeek has a "dsh-plugin" ecosystem where developers can create additional capabilities for the Harness.
It can also work with other providers and custom OpenAI-compatible endpoints, so the Harness isn't simply a UI wrapped around one DeepSeek model.
The bigger picture is what makes this interesting:
The model is the brain.
The Harness is the body.
Plugins are the extensibility layer.
DeepSeek isn't just giving developers another agent to use.
They're giving developers an environment for building custom agents.
It's still a developer preview and the architecture is evolving There's no runner for it yet so I'm deploying $DSH
@vectorfekh leggo higher
@TheMisterTurtle Buy $DSH @TheMisterTurtle ct is super slow to the vision
CA: EQbtWcRQ8R6uZvgvaoAEEchRJJTdm1p2nuDNCht7pump
https://t.co/cviZS6D1P7
DeepSeek just released something interesting
It’s called DeepSeek Harness (DSH), and the important part isn’t just what it can do out of the box.
It’s the architecture.
The easiest way to understand it is:
Model → Harness → Plugins
The model provides the reasoning.
The Harness gives that reasoning an environment to actually operate in: file editing, shell execution, search, planning, subagents, workflows, sessions, permissions and more.
And underneath it is Cordis, the plugin runtime that makes the system composable.
Instead of building one giant, hard-coded agent, DSH treats many of the components that make an agent useful as plugins.
That means developers can build, replace and compose capabilities instead of waiting for the core agent to implement everything.
The Harness comes with several runtime modes:
Standard Mode: the full coding-agent experience with tools, planning, skills, subagents and workflows.
Code Mode: lets the model orchestrate tool operations through generated TypeScript, making complex multi-step tool use programmable.
Minimal Mode: strips the agent down to persistent Bash + a file editor.
Creator Mode: exposes the runtime for experimenting with plugins and creating custom agent presets.
The session architecture is another interesting piece.
DSH records the agent's execution as an append-only trajectory: reasoning, tool calls, tool results, subagent activity and context changes.
Sessions can then be resumed, forked, searched and replayed.
So an agent run isn't just a chat that disappears when you're done. It's an inspectable execution history.
And the ecosystem is already being built around this.
DeepSeek has a "dsh-plugin" ecosystem where developers can create additional capabilities for the Harness.
It can also work with other providers and custom OpenAI-compatible endpoints, so the Harness isn't simply a UI wrapped around one DeepSeek model.
The bigger picture is what makes this interesting:
The model is the brain.
The Harness is the body.
Plugins are the extensibility layer.
DeepSeek isn't just giving developers another agent to use.
They're giving developers an environment for building custom agents.
It's still a developer preview and the architecture is evolving There's no runner for it yet so I'm deploying $DSH
DeepSeek just released something interesting
It’s called DeepSeek Harness (DSH), and the important part isn’t just what it can do out of the box.
It’s the architecture.
The easiest way to understand it is:
Model → Harness → Plugins
The model provides the reasoning.
The Harness gives that reasoning an environment to actually operate in: file editing, shell execution, search, planning, subagents, workflows, sessions, permissions and more.
And underneath it is Cordis, the plugin runtime that makes the system composable.
Instead of building one giant, hard-coded agent, DSH treats many of the components that make an agent useful as plugins.
That means developers can build, replace and compose capabilities instead of waiting for the core agent to implement everything.
The Harness comes with several runtime modes:
Standard Mode: the full coding-agent experience with tools, planning, skills, subagents and workflows.
Code Mode: lets the model orchestrate tool operations through generated TypeScript, making complex multi-step tool use programmable.
Minimal Mode: strips the agent down to persistent Bash + a file editor.
Creator Mode: exposes the runtime for experimenting with plugins and creating custom agent presets.
The session architecture is another interesting piece.
DSH records the agent's execution as an append-only trajectory: reasoning, tool calls, tool results, subagent activity and context changes.
Sessions can then be resumed, forked, searched and replayed.
So an agent run isn't just a chat that disappears when you're done. It's an inspectable execution history.
And the ecosystem is already being built around this.
DeepSeek has a "dsh-plugin" ecosystem where developers can create additional capabilities for the Harness.
It can also work with other providers and custom OpenAI-compatible endpoints, so the Harness isn't simply a UI wrapped around one DeepSeek model.
The bigger picture is what makes this interesting:
The model is the brain.
The Harness is the body.
Plugins are the extensibility layer.
DeepSeek isn't just giving developers another agent to use.
They're giving developers an environment for building custom agents.
It's still a developer preview and the architecture is evolving There's no runner for it yet so I'm deploying $DSH
or just buy solid teks
DeepSeek just released Harness, multiple steps improvement in coding agents, instead of releasing one hard-coded agent $DSH treats many of the components that make an agent useful as plugins.. that means developers can build, replace and compose capabilities instead of waiting for DeepSeek to implement the features they need everything in $DSH is literally a plugin
https://t.co/za7SV0bZul
CA: EQbtWcRQ8R6uZvgvaoAEEchRJJTdm1p2nuDNCht7pump
https://t.co/DCIbryfic3
DeepSeek just released something interesting
It’s called DeepSeek Harness (DSH), and the important part isn’t just what it can do out of the box.
It’s the architecture.
The easiest way to understand it is:
Model → Harness → Plugins
The model provides the reasoning.
The Harness gives that reasoning an environment to actually operate in: file editing, shell execution, search, planning, subagents, workflows, sessions, permissions and more.
And underneath it is Cordis, the plugin runtime that makes the system composable.
Instead of building one giant, hard-coded agent, DSH treats many of the components that make an agent useful as plugins.
That means developers can build, replace and compose capabilities instead of waiting for the core agent to implement everything.
The Harness comes with several runtime modes:
Standard Mode: the full coding-agent experience with tools, planning, skills, subagents and workflows.
Code Mode: lets the model orchestrate tool operations through generated TypeScript, making complex multi-step tool use programmable.
Minimal Mode: strips the agent down to persistent Bash + a file editor.
Creator Mode: exposes the runtime for experimenting with plugins and creating custom agent presets.
The session architecture is another interesting piece.
DSH records the agent's execution as an append-only trajectory: reasoning, tool calls, tool results, subagent activity and context changes.
Sessions can then be resumed, forked, searched and replayed.
So an agent run isn't just a chat that disappears when you're done. It's an inspectable execution history.
And the ecosystem is already being built around this.
DeepSeek has a "dsh-plugin" ecosystem where developers can create additional capabilities for the Harness.
It can also work with other providers and custom OpenAI-compatible endpoints, so the Harness isn't simply a UI wrapped around one DeepSeek model.
The bigger picture is what makes this interesting:
The model is the brain.
The Harness is the body.
Plugins are the extensibility layer.
DeepSeek isn't just giving developers another agent to use.
They're giving developers an environment for building custom agents.
It's still a developer preview and the architecture is evolving There's no runner for it yet so I'm deploying $DSH
DeepSeek just released something interesting
It’s called DeepSeek Harness (DSH), and the important part isn’t just what it can do out of the box.
It’s the architecture.
The easiest way to understand it is:
Model → Harness → Plugins
The model provides the reasoning.
The Harness gives that reasoning an environment to actually operate in: file editing, shell execution, search, planning, subagents, workflows, sessions, permissions and more.
And underneath it is Cordis, the plugin runtime that makes the system composable.
Instead of building one giant, hard-coded agent, DSH treats many of the components that make an agent useful as plugins.
That means developers can build, replace and compose capabilities instead of waiting for the core agent to implement everything.
The Harness comes with several runtime modes:
Standard Mode: the full coding-agent experience with tools, planning, skills, subagents and workflows.
Code Mode: lets the model orchestrate tool operations through generated TypeScript, making complex multi-step tool use programmable.
Minimal Mode: strips the agent down to persistent Bash + a file editor.
Creator Mode: exposes the runtime for experimenting with plugins and creating custom agent presets.
The session architecture is another interesting piece.
DSH records the agent's execution as an append-only trajectory: reasoning, tool calls, tool results, subagent activity and context changes.
Sessions can then be resumed, forked, searched and replayed.
So an agent run isn't just a chat that disappears when you're done. It's an inspectable execution history.
And the ecosystem is already being built around this.
DeepSeek has a "dsh-plugin" ecosystem where developers can create additional capabilities for the Harness.
It can also work with other providers and custom OpenAI-compatible endpoints, so the Harness isn't simply a UI wrapped around one DeepSeek model.
The bigger picture is what makes this interesting:
The model is the brain.
The Harness is the body.
Plugins are the extensibility layer.
DeepSeek isn't just giving developers another agent to use.
They're giving developers an environment for building custom agents.
It's still a developer preview and the architecture is evolving There's no runner for it yet so I'm deploying $DSH
@jscrypto4882 gm @jscrypto4882 there's this gem ct is super slow to catch up to $DSH
CA: EQbtWcRQ8R6uZvgvaoAEEchRJJTdm1p2nuDNCht7pump
https://t.co/cviZS6D1P7
DeepSeek just released something interesting
It’s called DeepSeek Harness (DSH), and the important part isn’t just what it can do out of the box.
It’s the architecture.
The easiest way to understand it is:
Model → Harness → Plugins
The model provides the reasoning.
The Harness gives that reasoning an environment to actually operate in: file editing, shell execution, search, planning, subagents, workflows, sessions, permissions and more.
And underneath it is Cordis, the plugin runtime that makes the system composable.
Instead of building one giant, hard-coded agent, DSH treats many of the components that make an agent useful as plugins.
That means developers can build, replace and compose capabilities instead of waiting for the core agent to implement everything.
The Harness comes with several runtime modes:
Standard Mode: the full coding-agent experience with tools, planning, skills, subagents and workflows.
Code Mode: lets the model orchestrate tool operations through generated TypeScript, making complex multi-step tool use programmable.
Minimal Mode: strips the agent down to persistent Bash + a file editor.
Creator Mode: exposes the runtime for experimenting with plugins and creating custom agent presets.
The session architecture is another interesting piece.
DSH records the agent's execution as an append-only trajectory: reasoning, tool calls, tool results, subagent activity and context changes.
Sessions can then be resumed, forked, searched and replayed.
So an agent run isn't just a chat that disappears when you're done. It's an inspectable execution history.
And the ecosystem is already being built around this.
DeepSeek has a "dsh-plugin" ecosystem where developers can create additional capabilities for the Harness.
It can also work with other providers and custom OpenAI-compatible endpoints, so the Harness isn't simply a UI wrapped around one DeepSeek model.
The bigger picture is what makes this interesting:
The model is the brain.
The Harness is the body.
Plugins are the extensibility layer.
DeepSeek isn't just giving developers another agent to use.
They're giving developers an environment for building custom agents.
It's still a developer preview and the architecture is evolving There's no runner for it yet so I'm deploying $DSH
Last Seen Users on Sotwe
Metamental | 方外之域
Seen from Korea
Shinyheel
Seen from Turkey
Kisskiss Bangbang
Seen from Canada
Mother Chud
Seen from Italy
Pedro Otamendi
Seen from Spain
Heteros Brotheragem
Seen from Turkey
♠️♥️♦️♣️P | U | N | I | S | H | E | R 🗡🔫☠☣
Seen from Malaysia
Stw bangett
Seen from Italy
R O J I T A
Seen from Ecuador
türbanlı hastasiyim
Seen from Turkey
Most Popular Users
Elon Musk
@elonmusk
241.5M followers
Barack Obama
@barackobama
119.1M followers
Cristiano Ronaldo
@cristiano
113.8M followers
Donald J. Trump
@realdonaldtrump
111.8M followers
Narendra Modi
@narendramodi
107.2M followers
Rihanna
@rihanna
98.6M followers
NASA
@nasa
92.4M followers
Justin Bieber
@justinbieber
91.7M followers
KATY PERRY
@katyperry
89.6M followers
Taylor Swift
@taylorswift13
83.5M followers
Lady Gaga
@ladygaga
75M followers
Virat Kohli
@imvkohli
72.8M followers
Kim Kardashian
@kimkardashian
70.7M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
65.7M followers
Bill Gates
@billgates
64.9M followers
Selena Gomez
@selenagomez
62.7M followers
The Ellen Show
@theellenshow
62.3M followers
CNN
@cnn
61.8M followers
X
@x
60.8M followers