GE
Gemini handles this as a draft: Use the definition I wrote. Say if my note is incomplete. Do not browse for a different definition.
Gemini · Unverified · gemini.google · 15 Oct 2026
Loading the catalog…
Gemini handles this as a draft: Ask a question beside a note, with only the context you choose. Nothing is written back until you save specific turns.
The post says type /Agent on a line in an EtherPK document. Pick how much the model may read: nothing else, the outline, the document, linked pages, or the whole graph.
The post's worked example is Gemini through OpenRouter, with the key staying on the device. Set a spending limit on the key. The page says Anthropic, OpenAI, Google Gemini, OpenRouter, Ollama, and LM Studio are options.
The model can only read. Saved turns become ordinary outline blocks. Leave the rest unsaved.
These are editorial labels so you can compare playbooks. They are not measurements.
No. Status is unverified. The EtherPK post was checked on 17 Oct 2026. The prompt is an adaptation.
What the model is allowed to read, and which turns are saved. Unsaved chat stays out of the note.
EtherPK blog post dated 2026-10-15. The page returned HTTP 200 on 17 Oct 2026. It describes /Agent inside a markdown note, a choice of how much context is sent, a read-only model, and saving only selected turns. The screenshots are a Gemini chat. Prompt and stop rules are an AI Playbook Gems adaptation, not a vendor quote; unverified.
Machine-readable copy: JSON
Gemini handles this as a draft: Use the definition I wrote. Say if my note is incomplete. Do not browse for a different definition.
Gemini · Unverified · gemini.google · 15 Oct 2026
Gemini handles this as a draft: Claims in my draft with no footnote. Do not invent sources.
Gemini · Unverified · gemini.google · 13 Oct 2026
Gemini handles this as a draft: Data, method, and the limit the authors name. Do not add results that are not in the PDF.
Gemini · Unverified · gemini.google · 30 Sep 2026