The .human protocol
Be the dotHuman in the loop.
An open protocol for reliable AI-assisted work.
A tool-agnostic operating layer that lives alongside your project — giving an AI agent a durable place to inherit your working method, survive context resets, and accumulate lessons that compound over time.
This site and its handbook were built with the protocol.
$ tree .human
.human/
├─ Comprehension/
# re-grounds every fresh session
├─ Goals/
# dated work folders, portable WIP
├─ Evergreen/
# runbooks that outlive a goal
├─ Reports/
# operational records, on demand
└─ captainslog.md
# the append-only spine
$
What it does
Comprehension/
A briefing that re-grounds a fresh agent
Goals/
Dated work folders that survive context resets
Evergreen/
Runbooks & lessons that outlive any single goal
Reports/
Operational records, created when needed
captainslog.md
The append-only spine that ties every session together — each loop writes here, so lessons compound.
Proof and deeper case studies are next.
Field reports are live now — read the protocol in the wild →