context continuity engine
Your project remembers itself.
nodestone keeps the plan, the decisions and their reasons, and progress that is proven, not reported — so every session, every AI and every teammate starts from where the project really is.
You explain the architecture. You explain why you picked Postgres. You paste the same conventions. Next session — gone. Your AI is brilliant and amnesiac. nodestone fixes the amnesia, not the brilliance.
with nodestone
Describe the project to your AI in your own words. It plans with nodestone — phases, tasks, dependencies, definition of done — and asks only what you must decide.
Every decision keeps its reason. A new session, a new model or a new teammate learns why, not only what.
Work counts as done when its checks pass, not when someone says so. Reports come from the records — nobody writes a status report.
The records live on your disk or your own team server. The license is checked offline. No telemetry.
Claude, Cursor, OpenCode or any MCP client works with the plan directly; the CLI does everything without AI too.
setup
pip install nodestonenodestone initclaude mcp add nodestone -- nodestone mcpUnderstanding, execution, and memory — one cognitive loop. atlas maps what's there, deepstrain changes it, and nodestone makes sure none of that hard-won context evaporates when the session ends.
pro
free for one person · $80 per person a year
also from us
deepstrain+atlasexecution + intelligence
The agent that does the work and the analyzer that maps your code. Pair them with nodestone's memory for the full loop.