Obsidian · Claude Code · Codex
Agent-Ready Vault Kit
Stop re-explaining your project to every new agent session. A plain-Markdown Obsidian vault with one entry point, twelve operating rules, a retry budget and a mission ledger, ready to open in Obsidian.

The problem it addresses
- Every new Claude Code or Codex session starts cold, so you paste the same context again.
- Agents read too much and follow every link until the context window is full.
- Two agents edit the same note and one silently overwrites the other.
- A failing step is retried until your quota is gone.
- An agent says "done" and there is nothing to check.
The six ideas behind it
They are free to copy. The kit is these ideas already built and documented.
One entry file
START_HERE.md is the only file an agent reads first. It says what the vault is and which note to load for which task.
Load only what the task names
Read the project master note, then only the files it lists. Never follow links recursively.
A mission ledger
One note lists who is working on what right now. Agents are told to check it before they change a shared note and to add their line before they start.
A retry budget
After a fixed number of failed retries, the agent is told to stop and write a handoff: what is done, what is left, the last good checkpoint.
No "done" without evidence
Each mission links an evidence note that says what changed and how it was checked.
One copy of the rules
CLAUDE.md and AGENTS.md only point to the rules, so the instructions for different agents cannot drift apart.
What's inside
- START_HERE: the one file every agent is told to read first.
- Twelve operating rules, including human approval before anything public, paid or destructive.
- Retry budget and stop conditions with a clean handoff format.
- Mission ledger: who is touching what right now, for agents to check before they change shared files.
- Project masters, jobs, missions, evidence, handoffs and daily notes.
- Nine blank templates and copy-paste prompts to start, resume, review and close work.
- A worked example and an implementation guide.
Agent-Ready-Vault/ ├─ START_HERE.md ├─ AGENT_OPERATING_RULES.md ├─ RETRY_AND_STOP.md ├─ MISSION_LEDGER.md ├─ CLAUDE.md · AGENTS.md ├─ PROMPTS.md ├─ IMPLEMENTATION_GUIDE.md ├─ Projects/ · Jobs/ · Missions/ ├─ Evidence/ · Handoffs/ · Daily/ └─ Templates/ (9 blank templates)
Opened in Obsidian
Screenshots of the v1.0 ZIP opened as a vault, in Obsidian's graph view.


Questions
- What do I get?
- A ZIP with 28 Markdown files: the vault structure, the rules, nine blank templates, copy-paste prompts and a worked example (a small newsletter project) that shows every piece in use.
- What do I need?
- Obsidian, which is free. It is written for Claude Code and Codex; any agent that can read files in a folder can be pointed at it. Git is recommended but not required.
- Do I have to rebuild my existing vault?
- No. The implementation guide shows how to add the kit to a vault you already use.
- Does it need plugins or an account?
- No. It is plain Markdown. Nothing is installed and nothing is sent anywhere.
- Does it enforce the rules?
- No. The rules are instructions, not enforcement: they work as far as your agent follows them. Keep real access limits in your agent's own permission settings and review its changes before you rely on them.
- What can I do with it?
- Use it for yourself and inside your own business. Resale or redistribution of the kit is not allowed.
- How is payment handled?
- Checkout, payment and delivery are handled by Gumroad. You receive the download right after purchase.