OpenTacit
OpenTacit helps you use AI better.
Models can do much more than most people ask of them. That gap between what a model can do and what its users try is called capability overhang. It is growing with the extremely rapid improvements in what models can do.
Capability overhang persists because the best ways to work with AI aren't easily shared. Power users get far more out of the same model—around seven times more, by one measure. The best AI users know what context to hand it, which tools to give it, and which steps suit a particular job.
That knowledge is tacit: known to the most advanced users, but unavailable to everyone else. OpenTacit captures it automatically, updates it when conditions and models change, and shares it right at the moment it is most needed. It watches how you work with your AI tools, determines what works best, and collects it as reusable techniques in a shared playbook. Then it offers those techniques back to you when the work calls for them.
Use OpenTacit on your own and it shows you how you work with models and agents, carries your best techniques across every agent you use, maintains them, and retires the ones you or your models have outgrown.
Share it with colleagues and it does much more, surfacing the best techniques from their collective human intelligence. Over time your OpenTacit playbook describes your business: its people, its processes, and the way they work.
This guide covers OpenTacit day to day: how to set it up, how to work with suggestions in your AI sessions, and how to run the registry that your organization shares.
Where to start
- You are trying OpenTacit on your own. Start with Start on your own. Two commands, and it ends with a technique in your next session.
- Your organization already runs OpenTacit. Start with Join your organization's registry, then read Work with suggestions in your session.
- You set up OpenTacit for your organization. Start with Set up a registry.
- You are new to OpenTacit. Read OpenTacit at a glance first. It explains the ideas that the rest of the guide uses.
The structure of this guide
- Start on your own — one person, two commands, and what changes when colleagues join
- OpenTacit at a glance — the ideas: techniques, the playbook, evidence, and the loop
- Set up a registry — install and configure the shared service
- Join your organization's registry — connect your machine and your AI tools
- Connect a chat tool — add OpenTacit to Claude or ChatGPT as a connector, with no installation
- How OpenTacit behaves in your tools — what runs automatically on each surface, how to interact, and the privacy line
- Work with suggestions in your session — the ◆ OpenTacit block, the @tacit advisor, and the status line
- Search and review on demand — find techniques and check your work when you want to
- Contribute a technique — capture a technique that worked, so that colleagues can use it again
- Give feedback — how OpenTacit collects evidence, and how to record it yourself
- Get to know the dashboard — sign in, navigate, filter, and sort
- Explore outcomes — see whether the playbook works
- Browse the playbook — the map, the library, and the tag vocabulary
- Review drafts and flagged techniques — the queue of decisions that need a person
- Manage members — invite colleagues and manage access
- Share techniques between organizations — publish and subscribe to technique feeds
- Configure the registry — settings, sign-in, storage, and the weekly digest
- Troubleshooting — diagnose and fix common problems
- Command reference — a list of every
tacitcommand - Glossary — definitions of the terms that OpenTacit uses
Read this guide anywhere
This guide is part of the registry itself. In the dashboard, click the account button in the top-right corner. Then select User Guide. The same files are in the repository in docs/user-guide/. Thus you can also read them on GitHub or in your editor.