🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.3 (31.07.2026)(31.07.2026 um 17:04 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.4 (08.08.2026)(08.08.2026 um 05:26 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.5 (08.08.2026)(08.08.2026 um 18:39 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.6 (12.08.2026)(12.08.2026 um 10:47 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.7 (13.08.2026)(13.08.2026 um 10:36 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.9 (18.08.2026)(18.08.2026 um 16:09 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.8 (20.08.2026)(20.08.2026 um 07:43 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.10 (20.08.2026)(20.08.2026 um 11:58 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.11 (23.08.2026)(23.08.2026 um 19:39 Uhr)
💾 DownloadsGitHub Release: Hmbown/Codewhale v0.9.12 (05.09.2026)(05.09.2026 um 10:21 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.3 (31.07.2026)(31.07.2026 um 17:04 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.4 (08.08.2026)(08.08.2026 um 05:26 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.5 (08.08.2026)(08.08.2026 um 18:39 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.6 (12.08.2026)(12.08.2026 um 10:47 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.7 (13.08.2026)(13.08.2026 um 10:36 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.9 (18.08.2026)(18.08.2026 um 16:09 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.8 (20.08.2026)(20.08.2026 um 07:43 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.10 (20.08.2026)(20.08.2026 um 11:58 Uhr)
🔧 AI Nachrichten GitHub Release: Hmbown/Codewhale v0.9.11 (23.08.2026)(23.08.2026 um 19:39 Uhr)
💾 DownloadsGitHub Release: Hmbown/Codewhale v0.9.12 (05.09.2026)(05.09.2026 um 10:21 Uhr)

26 🕛 kürzlich 5 Min Lesezeit
0

I sold you on /scratchpad. Then I migrated to /note.

↗ Quelle (dev.to)
🗣️ Stimme:
📑 Inhaltsübersicht

This is the third post in a series that started with . Two months ago I laid out the four-skill loop: /start-issue writes a /scratchpad, /tackle-scratchpad-block executes one step at a time, status flags transition from pending to done, and /finish-issue reads what shipped and writes the PR title + description.






Why I changed my mind



The driver was token consumption. My employer gave each team dashboards of our AI tool usage and I was always the top consumer on mine — fairly high on the department chart too. That's when I started to suspect it was not what I was building but how I was working with the tools.



/scratchpad's JSON step block and /tackle-scratchpad-block chain meant every task paid the same ceremony tax: status flags to transition, JSON to validate, a fenced block to parse. Useful scaffolding for multi-hour work. Overhead I was paying on every session. If my process was the reason I topped the dashboard, revisiting it was the cheapest experiment I could run.



Once I started looking, the redundancy wasn't only scratchpad-vs-note. The templates the composite skills were generating — what /start-issue, /start-side-quest, /tackle-pr-comment, and /finish-issue wrote into their output documents — carried weight they didn't need. A few examples from the audit that led to landed, Claude had gotten noticeably better at multi-step self-organization across an issue. I could lower my guard a bit. The control-freak posture I had started with was no longer needed. If I wanted an explicit checkpoint to manually review and commit, I could simply add a step in the generated /note.



Parallelism shifted too. Claude gained the ability to fan out workers and run agents in parallel within a single session, which meant on issues with independent pieces Claude could organize its own throughput faster than I could shepherd it through the /tackle-scratchpad-block chain. The gates that had once added safety started adding latency. By gating with /tackle-scratchpad-block, I was slowing Claude down by asking for more control.






What /note looks like



A /scratchpad's ## Implementation Plan section is a fenced JSON block with status fields, dependency arrays, and task lists:




CODE
{
"finish_issue_on_complete": false,
"steps": [
{
"id": "S001",
"title": "Swap the token library",
"status": "pending",
"done_when": "Old lib removed from package.json, new one imported and passing existing tests",
"depends_on": [],
"files": ["package.json", "src/auth/token.ts"],
"tasks": [
"npm install new-token-lib, npm remove old-token-lib",
"Update src/auth/token.ts imports",
"Run the token test suite"
]
},
{
"id": "S002",
"title": "Update callers to match new token API",
"status": "pending",
"depends_on": ["S001"],
"files": ["src/middleware/auth.ts", "src/routes/login.ts"],
"tasks": [
"Update verifyToken() call sites to new API shape",
"Run full test suite"
]
}
]
}






A /note's ## Plan section is the same information without the scaffolding:




CODE
1. Swap the token library — npm install new-token-lib, update imports, run the token test suite
2. Update callers to match the new token API — verifyToken() call sites, full test suite






No fenced JSON block. No status: pending. The plan says what to do and in what order; the LLM organizes its own execution in-session. The hard stops from the original workflow are still there: I read the plan before saying "go ahead", I review the diff, I commit when I'm satisfied. The commit lands at the end of the note, not after every step. If I want an earlier checkpoint, I add an explicit gate in the plan.



. The plan was a note this time.

Vollständiger Original-Bericht
Ausführliche Details, Code-Beispiele & Hersteller-Stellungnahme auf dev.to.
↗ Original-Artikel auf dev.to lesen
Wie bewertest du diesen Beitrag?
1 Klick Feedback
Teilen mit Netzwerk & Team:

Community-Analysen & Experten-Meinungen 0

Verfasse deine eigene Analyse, teile Workarounds oder diskutiere diesen Vorfall im Blog.
Noch keine Community-Analyse verfasst. Markiere einen Textabschnitt oder klicke oben auf Eigene Analyse verfassen“!
Community Pulse: Relevanz-Einschätzung
1 Klick Experten-Votum
🔴 Akute Relevanz 48%
🟡 In Evaluierung 22%
🟢 Keine Auswirkung 11%
Spannende Innovation 19%
Verwandte Story-Cluster & Quellen (Vektor-KI)
Port 8095 Engine
15 Quellen
GitHub Release: Hmbown/Codewhale v0.9.3 (31.07.2026)
Ähnliche Beiträge
🔍 Verwandte News

Auch interessante Nachrichten I sold you on /scratchpad. Then I migrated to /note.

Thematisch verwandte Begriffe: sold, scratchpad, Then, migrated · 6 Treffer

Laden...

Videos werden geladen ...

Laden...

Beiträge werden geladen ...

Laden...

Videos werden geladen ...

Laden...

Beiträge werden geladen ...

Laden...

Videos werden geladen ...

Laden...

Beiträge werden geladen ...

Laden...

Videos werden geladen ...

Laden...

Beiträge werden geladen ...

Laden...

Videos werden geladen ...