@witcheer

community @NousResearch · ex @KPMG

 Open Source
buttondown.com/witcheer
Joined October 2021
  • Tweets21K
  • Following1.5K
  • Followers22.6K
1.6KMedia
Hermes Agent gives you control
kache
@yacineMTBOct 24
Person who is reverse engineering his tv gets banned from OpenAI You aren't allowed to stop your tv from spying on your family. You have to ask for permission Or, maybe, maybe we accelerate
59
59
8
1,552
77,231
Hermes can remind you about those little and not so little things that are of importance in your day to day life 🪽
witcheer
@witcheerOct 23
forget to water the plants or take the bins out? your Hermes Agent can remind you, and keep reminding you until you've done it :) it asks what, when and where, then schedules a reminder that repeats until you reply. copy / paste prompt inn the first comment.
2
9
1,563
Yesterday AI was used to uncover major vulnerabilities in the infrastructure of the internet. It's yet another example of how profoundly AI is reshaping the web. But what emerges on the other side is very much up for grabs. We were lucky enough to recently introduce two heroes of the Rebel Alliance @eastdakota, CEO & cofounder of @Cloudflare @dillonrolnick, CEO & cofounder of @NousResearch, the creators of Hermes, the largest open source agent harness in the world. Together they discussed who gets to build, control, and benefit from an internet run by AI It's worth a listen if you care about the future of the web. 0:00 Preview 1:41 From a Discord Group to an AI Lab 8:45 Can Anyone Actually Pause AI? 12:59 Would You Trust an Agent With Your Inbox? 14:58 Why Matthew Wouldn’t Trade Places With Sam or Dario 20:28 The Dead Internet Theory Was Wrong 22:28 Agents Don’t Click Ads. Who Pays? 30:38 Where Should Your Agent Actually Run? 32:32 Why American Builders Are Using Chinese Models 38:37 What Happens When Our Agents Work Together?
25
43
17
374
81,616
Krea 3 from @krea_ai is now available in Nous Portal Fast mode does 2k edits in under 8 seconds, and Ultra is a higher quality tier outperforming Nano Banana Pro It supports precise controls: draw a box to create a bounded editing region leaving the rest of an image unchanged, or specify a hex code to apply precise color portal.nousresearch.com
Krea
@krea_aiOct 20
today, Krea 3 makes it to the top of image arenas. - "Fast" pushes the pareto curve of speed and quality. - "Ultra" ranks top-4, only surpassed by OpenAI models. the only company valued under $1T in the top-10! more below 👇
19
32
2
374
35,048
witcheer retweeted
Adolan
@adolandevOct 21
Replying to @NousResearch
@NousResearch Hermes picked up a chisel 🗿 Describe a prop or character: it models it with @tripoai, rigs it, makes it dance, and drops it into Godot, Unity or Unreal. Code and art, one agent 🎮 hermes plugins install Adolanium/hermes-tripo --enable github.com/Adolanium/hermes-…
3
4
1
32
1,199
witcheer retweeted
Dan Kornas
@DanKornasOct 20
Your agent doesn’t need every integration baked into the core. Hermes Official Plugins is the official plugin repo for Hermes Agent, for builders who want to add specific integrations without carrying every one in the core. It helps you install only the capability you need by exposing official plugins through the Hermes plugin catalog, where each entry is pinned to a reviewed commit. Key features: • One-line installs – install catalog plugins with `hermes plugins install <name>` • Reviewed commit pins – catalog entries point to the specific commit that was reviewed • Isolated plugin directories – each directory is a complete plugin; the layout includes manifests, dependencies, and tests • Practical integrations – the repo includes plugins such as Home Assistant, Spotify, and Snyk • Separate update cadence – plugin fixes can ship after review without waiting for the next Hermes core release It’s open-source (MIT license). Link in the reply 👇
6
11
40
2,993
Hermes merged 55 PRs on October 18 The ones worth knowing: • /usage in Desktop and the TUI now shows your provider account limits before you send the first message • Docker sandboxes are tighter: files can no longer carry setuid/setgid bits or file caps back to the host, and files written as root are handed back to your user at teardown • Plugins can now post to the current conversation and edit that message in place with ctx.post_message • Two new plugin hooks: post_compaction fires on every compaction path, and on_session_delete tells plugins about every session delete and archive • Plugin slash commands now receive the calling session on every surface • Clients that fall behind, reconnect or sit waiting now recover from the gateway instead of failing the turn • Unattended gateways no longer freeze and exit on a shell-hook consent prompt, and recorded consent registers the hook without a restart • Chat input stays in FIFO order, and session identity plus full history now survive compression, rewind and orphan reaping • Desktop session lists, resume and Bots groups stay stable across restarts • Stale dirty worktrees are archived before removal, and idle scratch clones are pruned again (one host had 100 GB held) • A wave of new catalog plugins, including WhatsApp voice, an Italian Desktop language pack, Hermes Gadget hardware, the Tsubasa model provider and Scalekit Here is the full list: DESKTOP IMPROVEMENTS (5 PRs) • Desktop/TUI /usage shows provider account limits before the first message • fix(desktop): New Bot's Capabilities tab loads when the name is typed later, and has room to use • fix(desktop): keep Desktop session lists, resume and Bots groups stable across restarts • fix(desktop): align main model settings controls • feat(plugin-catalog): add hermes-lang-it-full, an Italian Desktop language pack VOICE & REAL-TIME (3 PRs) • Clients that fall behind, reconnect or wait recover from the gateway instead of failing the turn • feat(plugin-catalog): add hermes-whatsapp-voice plugin • plugin-catalog: fish-audio v1.7.7 DEVELOPER TOOLS & PLUGINS (28 PRs) • Plugins get a post_compaction hook on every compaction path • feat(plugins): ctx.post_message: post to the current conversation and edit in place • Plugin slash commands receive the calling session on every surface • Plugins hear about every session delete and archive: on_session_delete hook • feat(plugins): public names for the skills tool's discovery and lookup helpers • feat(plugin-catalog): add scalekit • plugin-catalog: add completion-sound • catalog: codex-account-switch 1.2.3 (Hermes 0.22 support) • chore(plugin-catalog): re-pin hermes-windows-ops to a reachable commit • catalog: update Hermes Fleet Connector to 0.3.0 • feat(plugin-catalog): add edit-ledger • feat(plugin-catalog): add interim-ledger • catalog: add hermes-jetstream • feat(plugin-catalog): add gadget (Hermes Gadget hardware platform) • feat(catalog): add Tsubasa model-provider plugin • plugin-catalog: add hermes-seo • plugin-catalog: add offpeak-bekci (Ollama Cloud off-peak window + approval-gated work queue) • plugin-catalog: hermes-teammates 1.0.0 • plugin-catalog: bump hermes-vcc-engine to 0.4.1 • Update catalog entry: deepseek-search-fetch 1.0.6 (title, theme-safe banner, one known-issue note) • plugin-catalog: bump hermes-cron to 1.4.2 • chore(plugins): bump prompt-queue to latest commit • plugin-catalog: bump hermes-dreaming to 2.9.4 (the holographic provider's secret screen, a fact's origin) • chore(plugin-catalog): update Magi to v1.7.8 • plugin-catalog: openinvest re-pin to 0b9268b (0.25.0) • chore(plugins): pin birkin-mnemosyne to core 0.7.2 • docs(plugin-catalog): make Ace media discoverable by service name • docs(catalog): add Antenna banner image SECURITY & SECRETS (6 PRs) • Docker sandbox files can no longer carry setuid/setgid or file caps back to the host, persistent sandboxes included • Docker sandbox files written as root are handed back to your user at teardown • fix(secret-scope): the .env memo notices a same-size edit inside one timestamp tick • fix: profile-registry grants, hosted plugin route auth, and per-session Docker workspaces • fix(gateway): keep per-session settings and launch secrets that didn't stick • A gateway registers a shell hook once its consent is recorded, no restart INTEGRATIONS & CHAT (3 PRs) • fix(gateway): keep chat input in FIFO order and land Bot Chat deliveries whoever created the chat • fix(gateway): keep session identity and full history across compression, rewind and orphan reap • fix(gateway): serve goal-card criteria, subagent kill and steer, agent reactions and the connection card on the shared gateway STABILITY & FIXES (10 PRs) • Unattended gateway no longer freezes and exits 75 on a shell-hook consent prompt • fix(gateway): parked-profile ingress, downgrade relaunch, side-agent drain, idle exit, worker birth, Kanban ACK wait, log lock order • fix(gateway): a prune is not a permanent tombstone, and retirement is keyed by profile • fix(gateway): name an elevated gateway from its control pipe instead of OpenProcess • fix(update): a system-scope dashboard unit without root gets one exact restart command, not a polkit dump • fix(logging): the profile routing handler never builds a child handler under its own lock • Archive stale dirty worktrees before removing them, sweep every repo, ask on dirty exit • Idle scratch clones that share packs with a live repo are pruned again (100 GB held on one host) • Scratch prune logs one removal hint for root-owned entries instead of re-walking them hourly • test: the home-I/O guard exempts the PM generation dir around the running venv 55 new features and fixes xitter.cf/iamlukethedev/status/2…
Luke The Dev
@iamlukethedevOct 18
Hermes merged 113 PRs on October 17 The ones worth knowing: • Desktop now shows a lost turn with its Discard button when the gateway dies mid-turn, and you can steer or edit a queued prompt right from Desktop • patch and write_file return in milliseconds, and LSP diagnostics no longer hold up the write • Agents stop re-waiting on a quiet background process. Hermes tells them when it exits, and commands ending in a trailing & now run as tracked background jobs instead of being refused • LM Studio, Kimi and MiniMax streamed tool calls no longer shatter into broken calls • X Chat joins the plugin catalog as an official plugin, running as the account you connect • DeepSeek v4.1 Flash lands in the OpenRouter and Nous Portal pickers, and llama.cpp gets Nemotron 3.5 Bolt with DFlash drafts • Ctrl+D mid-turn in the CLI now detaches from the gateway instead of killing the run • Tighter secrets: frozen config secrets bind to their resolved endpoints, built-in keys stay off look-alike custom entries, and memory providers only serve their own profile • A big messaging sweep fixes WhatsApp, Telegram, Feishu, Discord voice and Bot Chat edge cases • Background one-shot calls stop paying for prompt-cache writes nobody reads • Slow session viewers no longer get dropped mid-reply on busy hosts • The plugin catalog grew a lot: Unraid, Nextcloud Talk, Grok Voice via hermes-xai, Jina, pay, personae and more Here is the full list: DESKTOP IMPROVEMENTS (8 PRs) • fix(desktop): show a lost turn and its Discard after the gateway dies mid-turn • fix: pre-upgrade sessions follow a moved custom endpoint; no setup card over a configured Desktop boot • feat(gateway): steer and edit a canonical queued prompt from Desktop • fix(gateway): send Ink and Desktop the turn's reasoning again • fix(desktop): list a chat the backend filed in Home even when it carries a cwd • fix(desktop): pinned removals, legacy-remote snapshot reads, insecure-origin dashboard writes • plugin-catalog: add quote-ask (Desktop: quote a selection into the prompt) • feat(plugin-catalog): add machine-spirit (community, desktop) VOICE & REAL-TIME (4 PRs) • feat(plugin-catalog): add hermes-xai (Grok Voice for Desktop + read-only xAI Collections search) • feat(plugin-catalog): add voice-ledger • LM Studio / Kimi / MiniMax streamed tool calls no longer shatter into broken calls • Slow session viewers no longer get dropped mid-reply on loaded hosts DEVELOPER TOOLS & PLUGINS (62 PRs) • patch/write_file return in milliseconds; LSP diagnostics no longer block the write • Agents stop re-waiting on a quiet background process; Hermes notifies them on exit instead • Foreground commands ending in a lone trailing & run as tracked background processes instead of being refused • fix(cli): Ctrl+D mid-turn detaches from the gateway instead of interrupting • fix(cli): Vertex /config label, /status id pin, and the classic CLI E2E comparisons • hermes update shows git's real fetch error, not only the From line • docs(api-server): what an Idempotency-Key covers after a session delete • Catalog installs of the Claude subscription and REA plugins get today's fixes • catalog: Claude subscription provider keeps its prompt cache across user turns after tool calls • catalog: aside-browser, duty-board, fish-audio 1.6.0, git-autosave, plan-mode 0.6.8, boardstate 1.5.2, limbic 0.6.4 • catalog: make-parts-now 1.0.1, blackbox • fix(catalog): update hermes-tailscale to 0.0.7 • fix(catalog): update hermes-resetwatch to 0.2.23 • chore(catalog): add a banner image to shodan • chore(catalog): add a banner image to openalex • chore(catalog): add a banner image to hermes-toolsmith • chore(catalog): add a banner image to hermes-terminal • chore(catalog): add a banner image to hermes-ssh • chore(catalog): add a banner image to hermes-session-diff • chore(catalog): add a banner image to hermes-rss • chore(catalog): add a banner image to hermes-office • chore(catalog): add a banner image to hermes-nous-prices • chore(catalog): add a banner image to hermes-ledgerline • chore(catalog): add a banner image to devin • chore(catalog): add a banner image to cloudflare-web-search • chore(catalog): add a banner image to clinepass • feat(plugin-catalog): add hermes-jina plugin • feat(plugin-catalog): add DeadBolt reviewed artifact writes • chore(plugin-catalog): update letsfg to 1.1.1 • Add catalog entry: deepseek-search-fetch • plugin-catalog: bump vector-platform to v0.5.3 • feat(plugin-catalog): add field-academy plugin • feat(catalog): add Unraid plugin • plugin-catalog: effortless-hermes v0.3.0 • Add usage-guard to the plugin catalog • feat(plugin-catalog): add stop-ledger • feat(plugin-catalog): add hermes-wslc-sandbox (WSL Containers terminal backend) • catalog: bump hydradb to 1.0.1 • feat(plugin-catalog): add comment-scraper • chore(catalog): pin scamgate to v0.3.0 • chore(catalog): holographic 0.17.3: which names are entities; delete, rename and split entities • fix(catalog): update Senler.io to 0.5.6 • Add tokstat to the plugin catalog • fix(plugin-catalog): update Model Router to v0.4.4 direct-Main policy • feat(plugin-catalog): add jp-reinfolib • plugin-catalog: bump hermes-lang-pt-br to f7b4357 (catalogs complete against main) • feat(plugin-catalog): add hermes-gensyn-ree • feat(plugin-catalog): add hermes-gensyn-delphi • catalog: bump iowap to 0.3.2 • feat(plugin-catalog): add bot-cabinet plugin • chore(plugin-catalog): repin kanban-gantt to 51d9e6a and pin 1.6.1 • catalog: add hermes-federated-search • plugin-catalog: add wallpaper-theme • feat(plugin-catalog): add Hermes Fleet Connector v0.1.5 • feat(plugin-catalog): add pay and pay-cli • feat(plugin-catalog): add personae • plugin-catalog: hermes-cloud-file-manager v0.4.0 • plugin-catalog: bump hermes-teammates to 0.1.3 • feat(plugin-catalog): add ample plugin • plugins: add hermes-artifacts private artifact library • feat(catalog): add hermes-muse companion assembly • plugin-catalog: bump hermes-lcm-x to 0.27.0 SECURITY & SECRETS (5 PRs) • fix(gateway): bind frozen config secrets to their resolved endpoints, not the whole policy • fix(providers): keep a built-in's key off a custom entry that shares its name • fix(memory): serve a memory provider only to the profile it was initialized for • fix(gateway): admit a user_id-routed sender on its routed profile • fix(gateway): don't send messaging users to /login for a host-side resolution failure INTEGRATIONS & CHAT (14 PRs) • X Chat joins the plugin catalog as an official plugin • X Chat catalog card: Hermes runs as the account you connect • feat: add Nextcloud Talk plugin catalog entry • deepseek/deepseek-v4.1-flash-1017 available in OpenRouter and Nous Portal pickers • feat(local-runtime): Nemotron 3.5 Bolt and DFlash drafts on llama.cpp b11671 • Platform plugins get public core helpers for authz, HTTP, setup and clarify (QQ Bot handoff) • refactor(platforms): core reads per-platform settings from each plugin instead of hand-kept tables • fix(gateway): a Bot Chat DM refused while another writer holds the chat is a definite no-send • fix(gateway): keep booting after a deleted Bot Chat's tombstones expire • fix(gateway): one set of local bots for groups and the roster, canonical group tab • fix(gateway): multiplexed WhatsApp TIME-WAIT, Feishu comment scope, Discord voice, webhook cron profile • fix(gateway): wake anchors never reach Telegram reply_to, heartbeats into GUI sessions, /retry admits before rewriting • fix(acp): named replies for unserved commands, provider-prefixed picks, overlay homes • fix(acp): map thought, plan, tool-state and image events on the gateway path STABILITY & FIXES (20 PRs) • test(windows): live Windows proof, plus a Windows packaging fix • fix(windows): user-usable dirs after an elevated run, named elevation/other-session refusals, lock-dir recovery • fix(state): concurrent startups on a zeroed state.db no longer report it read-only • TUI switcher-close tests stop failing at random on loaded CI runners • fix(sessions): catch a lagging state.db up once on browse instead of 503 • fix(gateway): live skill commands in Ink, clarify batches for local viewers, owner clarify.lock • fix(state): store a settled turn's reply once in its admission result • fix(gateway): session.usage also carries the nested usage view; deprecate the flat counters • fix(scratch): keep an idle entry a live Hermes-owned process works in; reap only orphans • fix(gateway): commands and one-shot output behave on the gateway as they did on main • fix(gateway): restore older local receipts and keep stored routes canonical • fix(api): keep API idempotency and image cleanup working across session delete • fix(gateway): log the cause of silent owner, storage and control-path failures • v0.22.2 installs fast-forward to main (record v0.22.2 as an ancestor) • fix(gateway): session writes from serve/dashboard reach the owner and fail by name • fix(gateway): adopt pre-update API session keys, keep pending clarify cards on re-entry • fix(gateway): reopen persisted sessions the gateway hasn't loaded, compressed-child resume, TUI switcher close • test(uninstall): control-socket uninstall test no longer races its own event-loop close • Browser tools no longer crash on deeply nested page data • Background one-shot calls no longer pay a prompt-cache write nobody reads 113 new features and fixes xitter.cf/iamlukethedev/status/2…
5
2
3
45
23,534
5 commands every Hermes user should know. These give you more control over Hermes while it's working, letting you change direction, add instructions, or kick off other tasks without waiting for everything to finish. • `/steer` : Guide the agent mid-task • `/queue` : Line up the next message • `/btw` : Ask a side question without interrupting • `/bg` : Kick off separate work in the background • `/busy` : Control how Hermes handles messages while working Hermes Desktop also lets you edit or delete queued messages before they're sent. @witcheer's eagle shows us how to put a couple of these to work.
witcheer
@witcheerOct 16
you can add to what your Hermes Agent is doing without waiting for it to finish, in two ways, in Hermes Desktop: - send a message while it works: it joins the current turn, and the answer takes it into account - press Cmd+Enter: the message waits in a queue above the composer and sends when the agent finishes queued messages can be edited or deleted before they go out.
5
5
106
8,083
witcheer retweeted
dmayhem93
@dmayhem93Oct 6
really happy to get this index out! huge thanks to @qorprate @yaboilyrical @JSupa15 @morganemoss @phucnhx for all the work and @daytonaio for supporting the eval runs! special shout out to gemini 3.8 flash for being our best answer leakage detector. not the job we gave it, but very helpful, thank you
Nous Research
@NousResearchOct 6
Introducing Hermes Index: an average of benchmarks measuring model performance in Hermes Agent. Our goal with Hermes Index is to better inform users choosing between models and to show labs how they perform in Hermes. portal.nousresearch.com/benc…
10
3
1
51
3,150
witcheer retweeted
Teknium 🪽
@TekniumOct 7
A lot of little things that help improve voice mode have been added to Hermes Agent this week!
54
60
11
802
31,031
is there something Hermes can’t do? 💯👀
witcheer
@witcheerOct 6
what to give your Hermes Agent first came up a lot this week. here are five of the jobs you asked about, each with a prompt to paste: (1) I'm job hunting. where do I start? give it a tracker file and a weekday search. it fills the file with listings that fit, then shows you the search job before creating it: ``` I'm job hunting for <role> roles in <place>. Make jobs.md with one row per job: company, role, link, status, next step, date added. Search for new listings that fit and add the best five. Then show me a cron job that repeats the search every weekday at 9am and reminds me of next steps that are due. Wait for my yes before creating it. ``` (2) can it do deep research? ask it to split the question between subagents. they research in parallel, and the report it writes links its sources: ``` Research <question>. Split it into three parts and give each part to a subagent working in parallel. Each one saves its sources with links to sources.md. Then write report.md: the short answer first, the evidence under it, a link on every claim, and a list of what you could not confirm. ``` (3) my notes are all over the place. can it turn them into a plan? point it at the folder. it reads every note and writes the week's plan, with the note's name next to each item: ``` Read every note in <folder>. Write plan.md for this week: what is due and when, what is waiting on someone else, and the three things to do first. Put the note's file name next to each item. Don't change anything in <folder>. ``` (4) can it help me practise a language? yes. at the end of a practice it saves your level and the mistakes you repeat to memory, and the next session loads them: ``` I'm learning <language>, around <level>. Run a 10-minute practice with me: chat with me in <language> about my day, one question at a time, then correct my mistakes at the end. Save my level and the mistakes I repeat to memory, so next time we start from there. ``` (5) what's the first step for selling my old gear? start with the listings. it finds similar items for sale online and drafts a title, description and price range for each one, with the links it used: ``` I'm selling some used gear: <items and their condition>. For each item, find similar listings online and write a listing draft: title, description, condition notes, and a price range with the links you based it on. Save them to listings.txt. Don't sign in anywhere or post anything. ``` happy to answer any other questions in the comments :)
5
2
23
2,307
witcheer retweeted
Tony Simons
@tonysimonsOct 6
Hermes Agent gets a LOT more interesting when you stop asking “what can it do?” and start giving it actual jobs. Job hunting. Research. Planning. Language practice. Selling old gear. @witcheers just handed you 5 damn good starting points. 👇
witcheer
@witcheerOct 6
what to give your Hermes Agent first came up a lot this week. here are five of the jobs you asked about, each with a prompt to paste: (1) I'm job hunting. where do I start? give it a tracker file and a weekday search. it fills the file with listings that fit, then shows you the search job before creating it: ``` I'm job hunting for <role> roles in <place>. Make jobs.md with one row per job: company, role, link, status, next step, date added. Search for new listings that fit and add the best five. Then show me a cron job that repeats the search every weekday at 9am and reminds me of next steps that are due. Wait for my yes before creating it. ``` (2) can it do deep research? ask it to split the question between subagents. they research in parallel, and the report it writes links its sources: ``` Research <question>. Split it into three parts and give each part to a subagent working in parallel. Each one saves its sources with links to sources.md. Then write report.md: the short answer first, the evidence under it, a link on every claim, and a list of what you could not confirm. ``` (3) my notes are all over the place. can it turn them into a plan? point it at the folder. it reads every note and writes the week's plan, with the note's name next to each item: ``` Read every note in <folder>. Write plan.md for this week: what is due and when, what is waiting on someone else, and the three things to do first. Put the note's file name next to each item. Don't change anything in <folder>. ``` (4) can it help me practise a language? yes. at the end of a practice it saves your level and the mistakes you repeat to memory, and the next session loads them: ``` I'm learning <language>, around <level>. Run a 10-minute practice with me: chat with me in <language> about my day, one question at a time, then correct my mistakes at the end. Save my level and the mistakes I repeat to memory, so next time we start from there. ``` (5) what's the first step for selling my old gear? start with the listings. it finds similar items for sale online and drafts a title, description and price range for each one, with the links it used: ``` I'm selling some used gear: <items and their condition>. For each item, find similar listings online and write a listing draft: title, description, condition notes, and a price range with the links you based it on. Save them to listings.txt. Don't sign in anywhere or post anything. ``` happy to answer any other questions in the comments :)
5
1
16
2,019
Hermes users are seriously sleeping on this memory prompt. This is one you should be running immediately on every Hermes agent you have. It turns your last 30 days of chats into a memory audit. Hermes looks for preferences, corrections, and facts you’ve told it more than once, checks them against what it already has in memory, then surfaces only the gaps. It even gives you the original wording, the dates you said it, and the memory entry it would add. Then it stops. Nothing gets added until you decide what actually belongs there. Run this, review the gaps, and decide what should finally stick.
witcheer
@witcheerOct 4
what you have told your Hermes Agent more than once that it still has not saved to memory, in one memory-gaps.txt. it searches your past chats for what you repeated and checks each one against its memory. copy / paste prompt in the first comment.
6
12
2
165
13,928
witcheer retweeted
tonbi
@tonbistudioOct 5
It's now much easier (and quicker) to connect your Hermes Agent to Discord What was previously a pain of wrangling configurations and searching for fixes is now super easy with the Hermes gateway setup wizard I made a short video walking you through the process, check it out!
Teknium 🪽
@TekniumOct 4
Setting up your Hermes with Discord should now be a little less confusing or painful
8
5
62
4,940
MCPs are one of the most powerful ways to extend what Hermes can actually do. 💥 They give your agent access to outside tools, services, and data, including many of the things you already work with every day. And Hermes Desktop has made connecting them stupid simple. Search the catalog. Hit Install. Start using the new tools. This is a part of Hermes people should absolutely be exploring.
witcheer
@witcheerOct 2
how to add an MCP server to Hermes Agent in one click, from Hermes Desktop: 1. open Capabilities > Connectors 2. search the catalog, here for DeepWiki, which answers questions about any public GitHub repo 3. click Install, and the card lists the tools it brings 4. open a new chat and ask for it by name
5
14
2
104
9,275
Side project going nicely all automated by Hermes Agent
9
3
1
39
16,967
20+ Hermes videos every user should know about 🔥 Real setups. Business use cases. Skills. Multi-agent teams. Marketing. Sales. Research. And a whole lot more. Witcheer basically built a video roadmap for learning what people are actually doing with Hermes.
witcheer
@witcheerSep 28
here is a map of the Hermes Agent videos worth watching, sorted by what you want to do with it: (1) start here (2) running a business on it (3) real setups and use cases (4) skills (5) teams of agents: Kanban, Bot Mode, cron (6) by job: marketing, sales, support, research see thread below
1
16
169
15,909
10 Hermes prompts every user should have saved. 👇 > Teach Hermes how you actually work > Turn finished work into reusable skills > Audit and improve your agent setup > Automate a weekly review of your work > Build the right AGENTS.md for a repo > Create recurring briefings around your interests > Review and organize the skills you’ve accumulated > Audit the files influencing your agent > Have Hermes create its own reference manual > Catch documentation work hiding inside open PRs Witcheer put together all 10 as copy-paste prompts on one page. Seriously useful collection. Definitely worth saving.
witcheer
@witcheerSep 29
10 Hermes Agent prompts so far, all on one page: notwitcheer.github.io/hermes… paste each one into a fresh Hermes Agent chat: - make your agent know how you work - turn what you just did into a skill - audit its own setup - a Monday review that writes itself - the AGENTS.md for your repo - a daily brief on the topics you name - a skills inventory it writes for you - audit the files that shape it - its own one-page manual - a docs to-do list from your open PRs
5
27
238
19,482
Hermes just made adding MCPs ridiculously easy. 🔥 Browse the catalog. Pick what you want. Click Install. Witcheer shows it here with DeepWiki. A few clicks and your Hermes agent has a whole new set of tools to work with.
witcheer
@witcheerOct 2
how to add an MCP server to Hermes Agent in one click, from Hermes Desktop: 1. open Capabilities > Connectors 2. search the catalog, here for DeepWiki, which answers questions about any public GitHub repo 3. click Install, and the card lists the tools it brings 4. open a new chat and ask for it by name
6
10
1
115
10,472
made empty promises to ur hermes agent? well here’s a way to make it up to him and find out what needs to be completed banger prompt by @witcheer
witcheer
@witcheerOct 2
what you told your Hermes Agent you would do and have not done yet, in one open-loops.txt. it searches your chats from the last 30 days and puts a date on each line. anything a later chat shows as done stays out. copy / paste prompt in the first comment.
2
1
13
1,848
Load more