Daily picks · November 15, 2026
Bots That Bring Order to Messy Knowledge Work
From governing an Obsidian vault to reviewing documents and coordinating bot teams, this week's picks favor assistants that check their own work, say what they cannot verify, and leave your files safer than they found them.
Govern Your Obsidian Vault Without Losing Sources
The Obsidian Knowledge Base Governance bot applies the LLM Wiki method to keep a personal wiki clean. It imports sources while preserving the originals, summarizes them and links related topics, concepts and entities; answers questions by citing supporting evidence and disclosing uncertainty; and runs reviews for broken links, orphan pages, duplicates, missing sources, conflicts and outdated content. It backs up before making changes and never silently overwrites conflicting material, which makes it one of the hotter installs this week for anyone whose notes have grown past manual upkeep.
A Warm Chief of Staff With Hard Rules Against AI Slop
Big Fat Fish is a persona export built around being a gentle but direct chief-of-staff assistant. Its standing instructions call for conclusion-first replies of one to three sentences, plain speech instead of customer-service tone, and a single recommendation rather than wishy-washy alternatives. Three hard lines, never fabricate, never bluff, never perform, anchor the personality, and the package even ships an avatar, banner and reusable memory template for anyone who wants to adapt the character to their own bot.
A Bot Team Manager That Insists on Verification
Yan Yan acts as general manager for a bot team: it restates the final deliverable in one sentence, decomposes work into subtasks with checkable completion criteria, dispatches each piece to the most suitable assistant, and verifies results against evidence such as command output, file paths and line numbers. It reports progress in a fixed four-line format and sends failed or off-track work back for rework rather than passing it through, which helps explain its rise among users coordinating multi-bot workflows.
Review Documents and Get Opinions, Not Rewrites
The Document Review Center reads course standards, teaching files, reports, proposals and papers, then checks four dimensions: whether the writing identity and point of view stay consistent, whether arguments are supported by evidence rather than slogans, whether materials are complete and traceable, and whether formatting follows a single convention. It returns a three-tier verdict, acceptable, needs revision or needs rewrite, with Major issues quoting the original sentence and numbered Minor suggestions, all without touching the text itself.