Skip to content
Mnema docs

Dev mode

Dev mode is what a Mnema workspace becomes when you connect coding agents to it. It turns the workspace into a control plane for agent work: every Claude Code session is captured with its real token cost, the files it touched, and the git branch it ran on — then rolled up into cost dashboards and optimization findings.

You turn it on per workspace in Settings → Developer (or when you first create a workspace, pick the developer template). Dev mode means one thing: this workspace has coding agents connected.

SurfaceWhat it doesDocs
SessionsEvery agent session with developer, model, duration, cost, git branch/commit and files touchedSessions
CostReal per-model token spend, budgets, and 50/80/100% alertsCost intelligence
OptimizeROI-ranked findings — stalls, retries, cost overruns, context bloatOptimize
Projects & boardKanban, projects and sprints (generalist — available in every workspace)Projects & board

Session, cost and git capture come from a small local hook you install once. See Connect your coding agent — it names exactly what data leaves your machine (metadata and file paths only, never file contents) and how to uninstall.

The whole dev workflow is free — on Mnema Cloud’s free tier and on the self-hosted community edition alike. Capturing your own agent work, your own cost, your own board costs us nothing per user, so it costs you nothing.

What’s paid is reasoning across that work — the knowledge graph over sessions, issues and decisions, cross-repo rollups, and org-wide analytics. In short: capturing your work is free; understanding it across time, people and repos is the paid layer.

See Self-hosting for exactly which dev features run with no licence.