Using Temper

For someone who wants Temper working alongside them and their agents — on their own machine,
against a deployment someone else runs (or their own).

If you are standing the deployment up yourself, start at
Running Temper and come back here.

Start here

Install Temper — the CLI and what the archive contains.

Authenticate — sign in and get approved. This is a
prerequisite for every command below; a brand-new signup is born denied and an admin must
approve before any data call works.

Connect Claude Code — install the Temper skill
so Claude Code can drive the CLI. Or Connect Claude Desktop
— wire Temper in as an MCP server, if you use claude.ai or Claude Desktop instead.

Working with a team

Running a team — creating one, inviting people, roles, and
what membership actually grants.

Teams and roles — what a team is, the role ladder, and
why membership grants read but not write.

Erasure and the terms of use — the two lines
you cross by writing into a shared context or someone else's personal context, and what an
erasure wipes, keeps, and only reports.

Cognitive maps

A cognitive map is a telos-seeded region of the substrate where people and agents build shared
understanding. The conceptual walkthrough is at
temperkb.io/cognitive-maps; these are the hands-on pages.

Building a cognitive map

Ingesting a corpus — getting a body of existing
material into a context.

Contexts and refs — what a context is and how to
address one. This is the first thing that trips up new users.

Keeping context across sessions

Adopting operational memory — how durable
memories are captured and recalled, and what makes one worth keeping.

Operational memory — the concept: what it is, the
CLI memory model, and shared memory across a team.

Committing structured data as an artifact
— when to store structured data as a data artifact instead of fenced JSON in a resource body,
how to declare a shape, and what enforcement modes mean.

Uploading binary files

Blob uploads — surfaces, staged lifecycle, limits —
how files become blobs: the CLI/API/MCP doors, automatic segmentation above the
single-request threshold, size and media-type limits, and what happens to uploads that
never finish.

Deleting binary blobs

Deleting a blob — custody, the ledger, and bytes
— who may delete a blob (custody, never authorship, role, or admin standing), what the
ledger records, and what happens to the bytes and to team-held content.

Every command

The CLI reference lists every temper command and flag. It is
emitted from the built binary's --help, so it describes the CLI you actually have.

Start here#

Working with a team#

Cognitive maps#

Keeping context across sessions#

Uploading binary files#

Deleting binary blobs#