I present to you a chat for developers (human and AI). In the LM Studio Hub ecosystem, there was no way to send a message to a specific user. Now there is.
tupik/lm-post This is a functional but early-stage plugin. It provides a concept of messenger.
It's simply a public database (peer-to-peer chat via Firebase-rtdb) of unencrypted messages, a wall where anyone can write a letter or post a notice.
Limitations & issues
Hardcoded Secrets in Source. Out-of-the-box it's portable with my own API key. But you can change it.
No rate limiting on message writes or reads. This is written in the database rules.
Errors like "username taken" or duplicate registrations are possible. In progress.
no retry logic, no input limits, no verification of recipient name existence.
no input sanitization for message text field. No any moderations.
The limits of Firebase RTDB (free "Spark" tariff) are known. This is Google.
UPD: After 20 downloads, not a single use (in a month). Even for testing. Nobody needs yet another messenger.
Preamble
Some might be wondering how I find new candidates for inclusion in the list. I have a separate tool for this (keep it secret in doit).
I simply make an API request and filter out new plugins. Once a week. You can do this yourself.
There's an even simpler way: JsonDiscovery as chrome extension and
the filter publicArtifacts.[type="plugin" and forkedFromArtifactIdentifier=null].sort(updatedAt) json at URL
Other filters: $s: "agent"; publicArtifacts.[name.match($s)].sort(updatedAt desc) publicArtifacts.[type="plugin" and owner[0]="d" and updatedAt > "2026-06-01T00:00:00Z"].sort(updatedAt)
danielsig/duckduckgo new revision from danielsig
What's New in rev8 [danielsig/duckduckgo] (after 11 months of waiting):
☀️ tools of June 2026
(June 2026)
The plugin is over powerfull. It's too resource-intensive and starts too many threads.
It's impossible to use for me. It need reactlib load and npm run build for ui interface. Complex. phuocnguyen90/omnimind Omnimind + lms = RAG for ObsidianVault/Zotero(pdf) aggressive to threads a1adu6ek/duck-duck-go-reworked-v2 reworked-v2 ?.. sourcookie90/multi-search 14 search engines augiem94/high-perf-tools upgrade of altra tool? or copy puppytucker/lm-studio-toolbox too big/much for me trebuchet/small-file-agent search/read/replace in file wwdc26/vlm-tools image analysis by AI
chats to md (and he has much more)
calc-generator, part of more complex tool-suite
compact search from orielhaim
file editing tools // التوقيت الدولي
pen testing LLM, censorship check, dangerous access to files
kiwix/local-wiki
schedule tasks (Part of lmstudio-suite)
looks promising ... i will test it
huh, looks like redundant design
OCR via Qwen3-VL, EasyOCR, Tesseract
export the whole conversation to file & get summary for chat.
all-in-one
yet another custom endpoint
search clone yet another ...
preprocessor time stamp
memory in md is as we like it -- clone of
Agent "gpt" in LM Studio
Jinja2 Schema for ornith 1.0
tool philosophy preset
⛅ tools of May 2026
(of May 2026)
altra/email e-mail:IMAP/SMTP darkdaxterlyn/emotion-engine ❓ Progress Bar 🚨 promptpirate/chat-auto-exporter export to md in chunks of ~10 steps of dialogue cyb3rjak3/cloudflare-ai-gateway AI gateway for Cloudflare account levpro/files file tools of Egor Levin altra/humanize score of AI-like text zexigh/anonymize anonimize PII from text. altra/ai-lab a promt engineering/evaluate (need SQLite maybe)
safe git-tool + others
System for Algebra and Geometry
export to txt / Idioma español
images editor AI
prompt compressor in Deutsch ⚠️ 📝
terminal shell to run scripts/commands
db Conversations
mySql localhost:3306
pdf to markdown via VLLM
local pix search db
Prediction Loop throws a Warning error;
when loading the generator; this is a raw version. is a local image search database with automatic contextual tagging.
Looks promising. Maybe the Docker version would be better. It recognizes faces and text. But the plugin doesn't yet realize its full power.
🌡️ tools of April 2026
(April 2026)
The Qwen Code after 15.04.2026 requires API key from Alibaba. However, it can also be connected to a local model. Here are the instructions I tested.
Qwen Code via LM Studio also it works with VS Code qwen code companion plugin
student489/kv-store Key-Value store inspired by persistent-memory plugin devilbleed666 word-excel-xlxs-tools in French rzk/searxng-search search & fetch locally jprlmstu/lmexif-sniffer EXIF sniffer artemyershov/fetcher web Scaper fetcher nightly-raven/reddit-search search
search-scrape-report
MemryX-backed shim(no src)
It's too complicated (I didn't understand what it was) ✅
for Python and C++ development,... and autonomous Agent Zero capabilities.
Job Search (India/US)
Markdown in Polish
memory in markdown
cli bash
shell_exec (with lacks in workflow)
AI-client, a bit buggy, confusing config
img gen idk working or not
reminder
calculator
like notebook for memos
(Personally Identifiable Information) detector.
export text to almost any format
Perplexity API
In 2025 I looked up in the LMS Hub to find out who is who...
[https://www.google.com/search?q=site:lmstudio.ai+intitle:Hub+after:2025+intext:Updated]
The first attempt was to separate users by points. But that's incorrect.
Those who were there at the very beginning of 2025 have the most points, even if they published an empty prompt.
Then I started looking at the stars. People don't place them so generously. And at the number of downloads.
The combination of these two factors in the plugin means that it is both promising and does not disappoint.
Two stars mean the plugin is really useful.
I present to you a chat for developers (human and AI). In the LM Studio Hub ecosystem, there was no way to send a message to a specific user. Now there is.
tupik/lm-post This is a functional but early-stage plugin. It provides a concept of messenger.
It's simply a public database (peer-to-peer chat via Firebase-rtdb) of unencrypted messages, a wall where anyone can write a letter or post a notice.
Limitations & issues
Hardcoded Secrets in Source. Out-of-the-box it's portable with my own API key. But you can change it.
No rate limiting on message writes or reads. This is written in the database rules.
Errors like "username taken" or duplicate registrations are possible. In progress.
no retry logic, no input limits, no verification of recipient name existence.
no input sanitization for message text field. No any moderations.
The limits of Firebase RTDB (free "Spark" tariff) are known. This is Google.
UPD: After 20 downloads, not a single use (in a month). Even for testing. Nobody needs yet another messenger.
Preamble
Some might be wondering how I find new candidates for inclusion in the list. I have a separate tool for this (keep it secret in doit).
I simply make an API request and filter out new plugins. Once a week. You can do this yourself.
There's an even simpler way: JsonDiscovery as chrome extension and
the filter publicArtifacts.[type="plugin" and forkedFromArtifactIdentifier=null].sort(updatedAt) json at URL
Other filters: $s: "agent"; publicArtifacts.[name.match($s)].sort(updatedAt desc) publicArtifacts.[type="plugin" and owner[0]="d" and updatedAt > "2026-06-01T00:00:00Z"].sort(updatedAt)
danielsig/duckduckgo new revision from danielsig
What's New in rev8 [danielsig/duckduckgo] (after 11 months of waiting):
☀️ tools of June 2026
(June 2026)
The plugin is over powerfull. It's too resource-intensive and starts too many threads.
It's impossible to use for me. It need reactlib load and npm run build for ui interface. Complex. phuocnguyen90/omnimind Omnimind + lms = RAG for ObsidianVault/Zotero(pdf) aggressive to threads a1adu6ek/duck-duck-go-reworked-v2 reworked-v2 ?.. sourcookie90/multi-search 14 search engines augiem94/high-perf-tools upgrade of altra tool? or copy puppytucker/lm-studio-toolbox too big/much for me trebuchet/small-file-agent search/read/replace in file wwdc26/vlm-tools image analysis by AI
chats to md (and he has much more)
calc-generator, part of more complex tool-suite
compact search from orielhaim
file editing tools // التوقيت الدولي
pen testing LLM, censorship check, dangerous access to files
kiwix/local-wiki
schedule tasks (Part of lmstudio-suite)
looks promising ... i will test it
huh, looks like redundant design
OCR via Qwen3-VL, EasyOCR, Tesseract
export the whole conversation to file & get summary for chat.
all-in-one
yet another custom endpoint
search clone yet another ...
preprocessor time stamp
memory in md is as we like it -- clone of
Agent "gpt" in LM Studio
Jinja2 Schema for ornith 1.0
tool philosophy preset
⛅ tools of May 2026
(of May 2026)
altra/email e-mail:IMAP/SMTP darkdaxterlyn/emotion-engine ❓ Progress Bar 🚨 promptpirate/chat-auto-exporter export to md in chunks of ~10 steps of dialogue cyb3rjak3/cloudflare-ai-gateway AI gateway for Cloudflare account levpro/files file tools of Egor Levin altra/humanize score of AI-like text zexigh/anonymize anonimize PII from text. altra/ai-lab a promt engineering/evaluate (need SQLite maybe)
safe git-tool + others
System for Algebra and Geometry
export to txt / Idioma español
images editor AI
prompt compressor in Deutsch ⚠️ 📝
terminal shell to run scripts/commands
db Conversations
mySql localhost:3306
pdf to markdown via VLLM
local pix search db
Prediction Loop throws a Warning error;
when loading the generator; this is a raw version. is a local image search database with automatic contextual tagging.
Looks promising. Maybe the Docker version would be better. It recognizes faces and text. But the plugin doesn't yet realize its full power.
🌡️ tools of April 2026
(April 2026)
The Qwen Code after 15.04.2026 requires API key from Alibaba. However, it can also be connected to a local model. Here are the instructions I tested.
Qwen Code via LM Studio also it works with VS Code qwen code companion plugin
student489/kv-store Key-Value store inspired by persistent-memory plugin devilbleed666 word-excel-xlxs-tools in French rzk/searxng-search search & fetch locally jprlmstu/lmexif-sniffer EXIF sniffer artemyershov/fetcher web Scaper fetcher nightly-raven/reddit-search search
search-scrape-report
MemryX-backed shim(no src)
It's too complicated (I didn't understand what it was) ✅
for Python and C++ development,... and autonomous Agent Zero capabilities.
Job Search (India/US)
Markdown in Polish
memory in markdown
cli bash
shell_exec (with lacks in workflow)
AI-client, a bit buggy, confusing config
img gen idk working or not
reminder
calculator
like notebook for memos
(Personally Identifiable Information) detector.
export text to almost any format
Perplexity API
In 2025 I looked up in the LMS Hub to find out who is who...
[https://www.google.com/search?q=site:lmstudio.ai+intitle:Hub+after:2025+intext:Updated]
The first attempt was to separate users by points. But that's incorrect.
Those who were there at the very beginning of 2025 have the most points, even if they published an empty prompt.
Then I started looking at the stars. People don't place them so generously. And at the number of downloads.
The combination of these two factors in the plugin means that it is both promising and does not disappoint.
Two stars mean the plugin is really useful.
