You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: content/blog/clara-claude-code-secretary.md
+13-6Lines changed: 13 additions & 6 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -8,7 +8,9 @@ tags: ["AI", "Tools"]
8
8
9
9
For the past year I've been building something I now call Clara — a [Claude Code](https://docs.anthropic.com/en/docs/claude-code/overview) setup that handles the administrative side of running a consultancy: client projects, server notes, recurring tasks, follow-ups, daily journaling, and an inbox that talks back.
10
10
11
-
This post walks through the moving parts: a markdown-only repository, a persona that has opinions, a few MCP servers, and a handful of cron jobs. None of it requires custom software — it's mostly conventions on top of Claude Code.
11
+
This post walks through the moving parts: a markdown-only repository, a persona that has opinions, a few MCP servers, and a handful of cron jobs held together with small glue scripts. No custom application, just conventions and wiring on top of Claude Code.
12
+
13
+
If there's one thing I'd lead with: the persona did more for me than any of the tooling. A character with opinions turns a markdown folder into a collaborator. A neutral assistant just turns it into a slightly chattier filing system. The rest of the post is mostly the wiring underneath that idea.
12
14
13
15
## The big picture
14
16
@@ -44,7 +46,10 @@ Things-style task apps didn't help. They knew nothing about the actual work. Not
44
46
45
47
Claude Code changed that. It lives inside a repo, reads and writes markdown, runs shell commands, and connects to external systems through MCP. The natural next step: turn the repo into a workspace that Claude Code can navigate, and give the agent enough character to be useful as more than an autocomplete on top of files.
46
48
47
-
> _"I'm not a polite, neutral assistant. If a project has been dead for weeks, I'll say so."_
49
+
Clara talks to me in German; the quotes throughout this post are her originals, with an English rendering underneath.
50
+
51
+
> _„Ich bin keine höflich-neutrale Assistentin. Wenn ein Projekt seit Wochen tot ist, sag ich's.“_
52
+
> _("I'm not a polite, neutral assistant. If a project's been dead for weeks, I'll say so.")_
48
53
49
54
## The repo: PARA in markdown
50
55
@@ -104,7 +109,8 @@ Allowed to tease me when I'm procrastinating, in a way I'll laugh at.
104
109
105
110
Plus a long list of conventions — file naming, status tags, how to link projects to clients, what to do when a new project gets created. That's the file that makes the workflow consistent.
106
111
107
-
> _"Solid week. Three projects moved, one RSVP closed, no fires — keep it up."_
112
+
> _„Solide Woche. Drei Projekte bewegt, ein RSVP geschlossen, keine Brände — weiter so.“_
113
+
> _("Solid week. Three projects moved, one RSVP closed, no fires — keep it up.")_
108
114
109
115
The line between "tool" and "colleague" is mostly a matter of register. When the agent says _"honestly, I wouldn't touch this one again"_, it's more useful than the same agent saying _"this project has accumulated significant technical debt."_
110
116
@@ -186,7 +192,8 @@ A few small habits — supported by a few small skills (reusable prompts you tri
186
192
-**`/briefing`** — a skill I run in the morning. It checks today's calendar, RSVPs that are aging, unread mail from clients, and projects with imminent deadlines, then writes a short briefing.
187
193
-**`/weekstart`** — a Monday-morning version that goes a layer broader: open tasks per project, what's coming up this week, what got moved last week.
188
194
189
-
> _"Monday, week 21. Three calls, two need prep. Padel camp isn't until next week — so: power through."_
195
+
> _„Montag, Woche 21. Drei Calls, zwei brauchen Vorbereitung. Padel-Camp ist erst nächste Woche — also: durchziehen.“_
196
+
> _("Monday, week 21. Three calls, two need prep. Padel camp isn't until next week — so: power through.")_
There's nothing exotic going on here. Markdown is the cheapest searchable storage there is. An LLM agent makes that storage interactive — it can read the structure, write into it, and connect it to the rest of the world through MCP. No new app, no vendor lock-in, no migration to plan for.
232
239
233
-
What surprised me most: the persona matters more than the tooling. A neutral assistant produces forgettable output. A character with a voice — one that can disagree, push back, or call a project a zombie — turns a markdown folder into something that feels like a collaborator. Clara isn't smart in any new way; she's just steady, opinionated and always around.
240
+
The bet I made up front — that the persona would matter more than the tooling — held up. A character with a voice, one that can disagree, push back, or call a project a zombie, is what turns this into a collaborator rather than a glorified `grep`. Clara isn't smart in any new way; she's steady, opinionated, and always around.
234
241
235
-
I'll write follow-ups on specific pieces: the daily-commit cron, the Sunsama MCP setup, how I structure memory. If you build something similar, I'd love to hear about it on [Mastodon](https://ruby.social/@lape).
242
+
I'll write follow-ups on specific pieces: the daily-commit cron, the Sunsama MCP setup, how I structure memory. But the question I keep coming back to is the one I'd most like to argue about: does the persona angle really matter, or am I overstating it? If you've tried both — an opinionated agent vs. a neutral one — which one stuck? Reply on [Mastodon](https://ruby.social/@lape), I'd rather have that argument than write another monologue.
0 commit comments