Who is Manence useful for?
Anyone who works with an AI agent on projects that last longer than one conversation:
a freelancer, a small team, an executive following several topics at once. It was
built on my three ventures, above. It asks you to be comfortable with a terminal and a
folder of files.
What do I need to use it?
git, bash, jq, python3 and Claude Code, on macOS, Linux or Windows with Git Bash. And
a real project for the first workstream.
What does it add if I already have an agent and skills?
An agent with skills knows how to do things. Manence gives it where to write and what
to keep: one workstream per project, a record, a decision log, a knowledge base kept
apart from drafts, and opening and closing steps. Your skills stay yours; they take
their place in the system.
Do I have to organize and file things myself?
No. Opening, keeping up to date, filing and distilling are the agent's moves, described
in its skills. You decide, you approve, you reread. You can also file things yourself
when you want to: these are your folders.
Can I take my projects to another agent?
Yes for the knowledge base, the log and the workstreams: markdown and git, nothing
else. Skills come along in the Agent Skills format, with a hookup to redo. Guardrails and
subagents need adapting for each tool. Verified in real sessions: Claude Code, OpenAI
Codex, Grok Build.
What does “my files stay on my machine” mean?
The system is a folder on your machine, versioned with git. There is no Manence server,
no account, no telemetry. What your agent sends to its model is a matter between the agent
and its vendor, not Manence.
Is Manence UI available, and is it required?
Not available: the repository is private, with no promised date. Not required: a
Manence system is complete without it. It reads what the system writes and writes nothing
itself; the same information is in the documents of the folder.
