
What happened
tanacode released version 1.1.0, adding checklists shared with Claude via an MCP server (tanacode-checklist) that persist through compaction and /clear, plus scheduled prompt sending at a chosen time.
Why it matters
Because the rules live in the list rather than the conversation, a long task handed to Claude is less likely to lose an instruction — such as checking everything before finishing — that compaction would otherwise drop.
What to watch
A scheduled prompt only sends while tanacode is running; if the app was closed or the Mac was asleep and it notices more than 5 minutes late, it keeps the prompt unsent and tells the user via a macOS notification instead.
WHO IT HITSPeople who hand long, multi-step work to Claude Code — such as developers running overnight tasks — get a place to record what should be verified before the work is called done.
Summaries like this, in your inbox every morning.
tanacode is a macOS app that sits alongside Claude Code and lets a person watch and check the work being done. The 1.1.0 release targets two everyday annoyances of delegating long tasks. The first is that an instruction given at the start — for example, confirming something before finishing — tends to fall away once the conversation is compacted. The second is that stopping for a question on every small judgment call leaves the work idle while the person is away.
The checklist feature addresses both by keeping lists separate from the conversation. Each list carries a name and a rule for how to use it, and Claude reads that description every time it opens the list. Claude records how it verified a check in the card's thread, and checks made by a person show in green while Claude's show in blue. A "questions to confirm" list lets Claude keep moving, writing down its judgment and reasoning so the person can review the cards later and correct them in the thread if needed. tanacode does not restrain Claude with hooks; how far to proceed is set by each instruction. The scheduled-send feature is meant for cases like reserving "continue the work" for when a five-hour window resets, or queuing next-morning instructions overnight.
The practical effect hinges on whether people keep the lists' usage rules up to date, since Claude follows what the list says. It may also depend on how reliably the MCP server's description and tool results keep Claude aware of human changes across a long session. The article says the checklists are meant as a place to leave what was delegated and what was verified, for work that spans compaction and /clear.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
ElevenLabs completed a $300 million employee tender at a $22 billion valuation, and co-founder Mati Staniszews…

On October 15, OpenAI announced the Rosalind Academic Plan, giving approved US university labs free access to…

A Qiita article compared automated red-teaming methods for LLM apps — GCG, PAIR, and TAP — using mock implemen…

Liquid AI released d1-ultra on October 15, 2026, an upgraded open-weights decision model that averaged 76.7 ac…

Circlia started a proof-of-concept for TASTE SIGNAL: visitors to Holland Village Cafe FUKUOKA taste cookies an…

In a webinar hosted by French Mathematical Society president Isabelle Gallagher, Fields Medal winner Hugo Dumi…
