The starter kit

What's inside

One SQL file for the database, four plain-text documents the assistant reads, three skills, and three scripts. Nothing runs on its own.

starter/
├── README.md                      install, daily use, troubleshooting
├── LICENSE                        MIT
├── schema/
│   └── obligations.sql            the whole database, one idempotent migration
├── ritual/
│   ├── SESSION_CONTEXT.md         the operating doctrine the assistant reads
│   ├── OPEN.md                    how a session opens
│   ├── CLOSE.md                   how a session closes
│   └── DOCTRINE_TEMPLATE.md       versioned laws learned from incidents
├── skills/
│   ├── session-ritual/SKILL.md
│   ├── approval-lattice/SKILL.md
│   └── capture/SKILL.md
└── scripts/
    ├── open.sql                   the three reads of the open
    ├── close.sql                  the close, as a fill-in template
    └── seed.sql                   optional example rows for day one

The database is four tables (tasks, decisions, session_events, session_handoffs), a trigger that gives every open item a check date, and two views: v_due_for_check for completeness and v_next_up for importance, with a why column that shows how each score was built.

Install in five steps

1. Create a free Postgres project 3 min

   Sign up for a free hosted Postgres such as Supabase, or use any Postgres 15 or newer you already run. Create a new project.

2. Run the schema 2 min

   Open the project's SQL editor. Paste the whole of this file and run it:

   starter/schema/obligations.sql

   It is safe to run twice. Optionally run starter/scripts/seed.sql as well, so the first open has six example tasks, one decision and one handoff to show.

3. Give the assistant its doctrine 3 min

   Read starter/ritual/SESSION_CONTEXT.md, adjust anything that does not fit, then paste it into the assistant's standing instructions: a Claude Project's instructions, a ChatGPT project or custom instructions, or a Gemini Gem. If it accepts files, add OPEN.md, CLOSE.md and the three skills as knowledge.

4. Open the first session 5 min

   Start a new chat in that project and type:

   Open the session

   Then run starter/scripts/open.sql in the SQL editor and paste the three results into the chat, or connect a database tool so the assistant runs it itself. It should handle every row it owns, list yours as one group, and offer two or three ways to spend the session.

5. Close it 2 min

   When you are done, type:

   Close the session

   The assistant fills in starter/scripts/close.sql: rows closed or re-dated, what shipped, what was ruled, and a handoff. Read it, then run it. It is one transaction, and the handoff is rejected if any of its four required fields is empty. The next open starts from it.

What this is not yet

Version one keeps every action inside a chat you are part of. The pieces below each add reach, which is exactly what the approval lattice exists to govern, so they come later, one at a time.

Each is described, tier by tier, in the guide.

Download

Or browse the files

• README.mdstart here
• LICENSEMIT
• schema/obligations.sqlthe database
• ritual/SESSION_CONTEXT.mdthe doctrine
• ritual/OPEN.mdthe open
• ritual/CLOSE.mdthe close
• ritual/DOCTRINE_TEMPLATE.mdlaws from incidents
• skills/session-ritual/SKILL.mdskill
• skills/approval-lattice/SKILL.mdskill
• skills/capture/SKILL.mdskill
• scripts/open.sqlthree reads
• scripts/close.sqlclose template
• scripts/seed.sqloptional examples

Questions

Does it need an API key?

No. The assistant you already pay for reads the context you paste into its instructions. The database is reached through its own SQL editor, or through a database tool you choose to connect.

Which assistant does it work with?

Any assistant with project-level or custom instructions it reads on every chat. Nothing in the kit is tied to one vendor.

Can I use my own schema?

Yes. Everything lands in public for simplicity; the header of obligations.sql shows the one line to add to put it in a schema of your own.