I use #Hermes as a personal AI chief of staff and automation layer. It manages my Todoist tasks, priorities, reminders, and follow-ups; researches current news, products, apps, sports, and investments; helps monitor and run my self-hosted servers, Docker apps, backups, and media stack; organizes my bookmarks, notes, and documents; and builds workflows across APIs, webhooks, MCP servers, and scheduled jobs. It can also process things like lottery tickets, track personal data in my systems, draft messages, and handle technical troubleshooting—so I can manage a lot of personal operations through a single chat.

Manus vs. HyperAgent vs. Tasklet?

Dream came true today. 🎉

Anthropic gave Pro subscribers a $100 usage credit — an apology for moving Claude Fable 5 fully behind usage credits. I don’t even use Fable 5, but here’s the thing: I run Claude in Hermes (my personal AI agent setup) via Pro’s OAuth, and it draws from a topped-up usage credit balance — not the regular Pro pool.

I’d been holding back on using it much because topping up out of pocket felt like an added expense on top of Pro. This $100 just removed that friction entirely — I finally get to use Claude in Hermes the way I actually wanted to.

Only wish: this were a monthly allowance instead of a one-time thing. If it were recurring, this would legitimately change how I use Pro long-term.

For now though — genuinely happy about this one. 🙂

#Claude #Anthropic #AI #Hermes

#Mattermost letting slash commands work in the browser but not on mobile is ridiculous. The #Hermes gateway/integration works, but the mobile app grabs anything starting with / before Hermes ever sees it. So the exact same command works on web, then dies on mobile. Same bot, same chat, different rules.

And the suggested workaround somehow makes it worse: add a leading space so it sends as plain text, except now you risk breaking the command for the integration on the other end. It’s a full custody battle over a single character. If Mattermost is going to hijack /, it should either support custom integrations properly everywhere or get out of the way.

#Grist shipped an official MCP server, which should be a big win.

But if you self-host, you need the full/Enterprise edition to actually use it.

So the people most likely to want MCP for private, local, agent-driven workflows are the ones gated out. What a waste of potential.

I keep seeing these references to MCP server of the #Grist changelog, but where is the full MCP server documentation? Where’s the official one, at least? #GetGrist

Claude Fable is not “slightly more expensive” — it’s a premium tier.

API pricing per 1M tokens:

Fable: $10 input / $50 output Opus: $5 input / $25 output Sonnet: $3 input / $15 output

So Fable is 2× Opus and about 3.33× Sonnet. For most coding/finance/reporting workflows, Sonnet still feels like the default; Opus for hard tasks; Fable only when it materially reduces turns or rework.

Jollibee just got demoted from the MSCI Standard Index to Small Cap. My portfolio got the same treatment. At least we’re consistent. 🍗📉

My average: ₱183. Current: ~₱134. That’s a ~27% discount I didn’t ask for.

MSCI effective date: June 1. So the index exclusionary is officially behind us. Now we wait for the chicken to fly again.