01 / Stack + real costs
A grounded map of the tools, model choices, and recurring costs behind a useful agent setup.
AO / FIELD MANUAL 01PDF · 21 pages · 9 parts
A practical operating system for AI agents
Stop improvising with agents. Start operating them.
The Agent Operator's Manual is a dense, candid field guide for technical founders, solo developers, agency operators, and serious Claude Code or Codex users who are tired of losing context, repeating mistakes, and accepting false-green work.
Get the manual · $9One-time purchase · start today
$ operator run --reliable
context loaded .......... OK
memory linked ........... OK
skill selected .......... OK
verification gate ....... PENDING
result: useful work, with evidence
Not prompt tricks. The operating layer around the model.
The problem
A capable model still produces unreliable work when the operating system is improvised.
01
Every new task starts with the same context dump.
02
The output looks finished until you test the edges.
03
You spend more on inference because the loop has no shape.
Inside the manual
Nine parts that turn agent usage into a repeatable practice.
02 / Operating contract
A compact contract that tells the agent how to work, when to ask, and what 'done' means.
03 / Laziness ladder
A practical way to trade speed against reliability without turning every task into a ceremony.
04 / Obsidian memory
A memory structure that keeps decisions, context, and lessons available after the chat disappears.
05 / Reusable skills
Turn good work into named procedures you can call again instead of re-explaining yourself.
06 / Verification gates
Catch false-green outputs with evidence checks, tests, and explicit handoff gates.
07 / Cheap infrastructure
Keep inference and execution costs sensible while the system earns its keep.
08 / Worked pipeline
Follow one task from brief to verified result, including the points where most agents quietly fail.
09 / 30-day path + templates
A month-long implementation path plus copy-paste templates for contracts, skills, memory, and checks.
What changes
You leave with a system you can actually put to work today.
Before: a blank chat, scattered notes, a vague request, and a confident-looking result.
After: a known stack, an operating contract, durable memory, reusable skills, and verification gates that make “done” observable.
Read it once. Copy the parts you need. Build from there.
The complete manual
The Agent Operator's Manual
21 pages. 9 parts. A practical agent-first operating system in a PDF you can keep beside your editor.
One-time access$9
- ✓ The complete 21-page PDF
- ✓ All 9 operating-system parts
- ✓ Copy-paste implementation templates
- ✓ Secure download after confirmed payment
Secure checkout · one-time purchase
This is a field manual, not a promise of effortless automation. It gives you the operating pieces and a path to assemble them.