Daisy' (Anthropic Claude Code engineer) workflow

2026-08-24 Mon 09:10:55

On <2026-08-23 Sun> I read in stumbled on this post - Anthropic: Please Have Daisy the CC Engineer Do a Video! : r/ClaudeAI.

In this week's newsletter, Anthropic included this:

"My daily driver currently looks like: two lead agents that keep each other accountable and restart the other if either fails. These delegate to tech lead or PM agents for the 8-10 projects I'm running at any one time, and each project has 5-10 IC [individual contributor] agents, generalists or specialists depending on the problem. Across all of these I'm still only doing 30-50 prompts per day, and my IC agents typically work autonomously for 2-3 days. About 60% of my interaction is with the leads, 35% with a project lead, and 5% is when something has gone off the rails. All of these agents communicate directly with the SendMessage tool."

– Daisy, Engineer on Claude Code

Turning it into mindmap:

                                      ╭─ keep each other accountable
                   ╭─ two lead agents ┼─ restart the other if either fails
                   │                  ╰─ delegate to ┬─ tech lead ── 8-10 projects...
                   │                                 ╰─ PM agents
                   │                                                 ╭─ . ┬─ generalists
« Daisy workflow » ┼─ 8-10 projects ── per project ── 5-10 IC agents ┤    ╰─ specialists
                   │                                                 ╰─ work autonomously ── for 2-3 days
                   │                        ╭─ 60% with leads
                   ├─ 30-50 prompts per day ┼─ 35% with project lead
                   │                        ╰─ 5% off rails
                   ╰─ agents communicate ── SendMessage tool

My thoughts:

Why two lead agents? Likely they both run "/loop 30m check projects" with offset, so one is checking status while another is ready for communication. This is where "60% with leads" coming from; sometimes they both are checking, and Daisy goes to project leads directly. Those two leads have permanent session names and never closing.

8-10 projects. I also have about the same number of big topics a week, often carried to other weeks. So this sounds like the right amount of parallelism. Also keeping 8-10 projects on track is within cognitive abilities of an average human.

Beyond the basics with Claude CodeYouTube

Comments lead to this video:

YT: Beyond the basics with Claude Code - YouTube
author: Daisy Hollman
date: <2026-05-22 Fri>

It breaks down to this mindmap:

                                          ╭─ Slack
                 ╭─ where work comes from ┼─ CI/CD ── no point to fix CI/CD manually
                 │                        ├─ Dashboards
                 │                        ╰─ Internal Docs
                 ├─ tip ── try doing ALL work from within CC
                 │         ╭─ code conventions are not in the model
                 ├─ limits ┼─ fine-tuning doesn't work well ── leads to more hallucinations
                 │         ╰─ in-context learning ── ICL ── general AI wins over specific AIs
                 │                                                         ╭─ rigid templates
                 │                     ╭─ compensates lack of intelligence ┼─ guardrails
                 │                     │                                   ├─ "don't let it touch X"
                 ├─ two kinds of tools ┤                                   ╰─ gets less useful as model improves
                 │                     │                           ╭─ more access
                 │                     ╰─ scales with intelligence ┼─ more control
                 │                                                 ├─ faster feedback loops
                 │                                                 ╰─ gets more useful as model improves
                 │                          ╭─ context window stayed at 1M max ── models are way better yoy
                 │                          ├─ can't dump the whole X into context ── X ┬─ wiki
                 ├─ context window is a box ┤                                           ╰─ codebase
                 │                          ├─ example ── try to run npm on Arduino
                 │                          ╰─ don't pay for what you don't use ── [TK: no slop skills]
                 │           ╭─ cached tokens are cheap ── new are expensive
                 │           ├─ not an LRU problem
                 ├─ KV cache ┼─ change in early instructions invalidates the cache ┬─ add new tool
                 │           │                                                     ╰─ change in CLAUDE.md
                 │           ╰─ put stable stuff at the front ── volatile at the end ── CC teams tries to solve this problem
                 ├─ Large-scale Software Engineering ── main question ── Does it scale?
                 │                           ╭─ when it's the right tool?
                 │                           ├─ CC has a shell ── Skill+CLI has less overhead
                 │                           │                                 ╭─ name
                 │                    ╭─ MCP ┼─ tools definitions sit in front ┼─ arguments
                 │                    │      │                                 ╰─ description
« Agentic flow » ┤                    │      │              ╭─ puts just name in the system prompt
                 │                    │      ╰─ tool search ┼─ kind-of lazy-loaded
                 │                    │                     ╰─ needs to be very specific ── e.g. slack
                 │                    │         ╭─ in-context
                 │                    │         ├─ one-line description always loaded
                 │                    │         ├─ full SKILL.md and assets loaded on-demand
                 ├─ Plugin primitives ┼─ Skills ┤                 ╭─ body is pay-per-use ── good
                 │                    │         ╰─ does it scale? ┼─ reliably loading requires a paragraph in description
                 │                    │                           ╰─ not possible to lazy-load sub-skills ┬─ no hierarchy yet
                 │                    │                                                                   ╰─ what if 100k skills?
                 │                    ├─ Hooks ┬─ runs code on events ── shell scripts
                 │                    │        ╰─ zero-overhead abstraction
                 │                    │         ╭─ out-of-context ── fundamental difference to a skill
                 │                    │         │               ╭─ named role
                 │                    │         ├─ more Claudes ┼─ own system prompt
                 │                    ╰─ Agents ┤               ╰─ own tool set
                 │                              ├─ spawned with a task ┬─ returns result
                 │                              │                      ╰─ its transcript doesn't come back
                 │                              ╰─ description still in-context ── what if 100k agents?
                 │                             ╭─ loads fully unconditionally ┬─ doesn't scale
                 │                ╭─ CLAUDE.md ┤                              ╰─ every plugin adds it
                 ├─ Doesn't scale ┤            ╰─ better inject via SessionStart hook
                 │                │         ╭─ low-quality ── low-cost ── useful, different contract
                 │                ╰─ Memory ┼─ model-curated, not human-curated
                 │                          ╰─ plugins are human-authored, human-reviewed
                 │                 ╭─ worktrees ┬─ one repo ── N claudes each owns its branch
                 │                 │            ╰─ better make it long-lived ── less overhead ── [TK: try kunchenguid/treehouse?]
                 ├─ Multi-clauding ┼─ asynchrony ┬─ walk away, let it do the work
                 │                 │             ╰─ /loop 10m
                 │                 ├─ parallelism ── multiple of asynchrony
                 │                 ╰─ lot's of context switching
                 ╰─ Agent-teams ┬─ teammates ── long-lived ── unlike agents
                                ╰─ SendMessage tool ── Claude's talking to each other

Follow-up questions:

Do I need claude-mem?

SafariTabs.app

2026-05-03 Sun 21:52:56

We need RSS for sharing abundant vibe-coded apps post by Matt Webb mentions Tabulator app, part of Wall of Apps by Matt Sephton. All the apps there share the same minimalist design, native Swift UI, tiny and, supposedly, fast. The downside is that they are paid.

I wanted to try Tabulator, as I use seal_safari_tabs.lua with Hammerspoon. So I built SafariTabs.app (using Claude Code, a project), source code is here – velppa/SafariTabs. Version v0.1.0.

Features:

Fuzzy-searching for tabs
Keyboard navigation
Switching to the selected tab
Close tab
Headless mode, "safaritabs:" URL scheme

Let's see if I'm going to use it, if yes – will think about publishing a release. If you want to build it – clone the repo and ask Claude Code to build it.

Trying Textpod once again

2026-04-14 Tue 22:49:44

On <2026-04-13 Mon> I installed Textpod on Mac Book Pro M1 using Cargo (Rust tooling).

Good:

Separate from notes, and from blog.
Managing ton of markdown files for small notes that I produce - huge overkill
I want something like this, write in org-mode, press a button (call elisp function) and heading is sent to website.
Open source, so I can fork and improve it for my needs.
- ☑ make a fork - ~/Developer/src/github.com/velppa/textpod
Can make it prettier.

Open questions:

☑ Does it have an API? Like I would select a Org text, convert it to Markdown and send to Textpod.
- use {M-x textpod-org-region-to-note} function
☑ How to increase font? Increased in the sources.
☑ Token access, for hosting on finita-la-comedia
☑ Internal Org Roam links are pasted as markdown files. So in theory, I could somehow expose them.

I ended up forking it to https://github.com/velppa/textpod and greatly improving it for my needs. You're reading notes served by Textpod v0.2.0. It is my project.

Notable changes:

Read-only mode with token authentication: The app now supports an optional --token flag that enables authentication. When a token is set, users without the token see a read-only interface where they can only search notes. Access is granted via a ?token=X query parameter (which sets a cookie) or a Bearer token in the Authorization header. The UI dynamically adjusts based on authentication status, hiding edit/delete controls and changing the editor placeholder for unauthenticated users.
Enhanced note management: Notes now have unique IDs based on timestamps, enabling direct access via /note/:id URLs. PUT method allows editing notes (upserting if not found).
Subpath support: to deploy on your domain under subpath, supply --base-path notes flag. Then notes will be under /notes subpath.
The app watches the notes file for external changes and automatically reloads.
The link archiving feature lets users prefix URLs with + to both raw urls and Markdown-formatted.
Special handling of links to .md files: They appear when exporting from Org-roam. These are internal links and Textpod now turns them into search by filename, effectively returning all mentions of this concept. Poor man zettelkasten import.
UI improvement: larger fonts, footer, better code snippets for dark mode, tags on the heading.
textpod.el: Companion Emacs package for quick sending notes to Textpod.

Later changes to setup will go into separate notes.

Vincent Van Gogh gallery

2026-04-06 Mon 13:02:44

I found on my external drive a collection of Vincent Van Gogh paintings, that I converted into desktop wallpapers using Backgroundifier app (archagon/backgroundifier-public) several years ago.

The collection is pretty big, 1380 paintings. To explore it in a better way than just going through files on a disk, I built an interactive gallery (a small project using Claude Code). It serves images directly from my external drive; also had to prepare thumbnails for faster loading.

Check it out here: https://hotter.myaddr.dev/van-gogh/, not bad for couple hours of work, don't you think?