TENET

Context that compounds.

Portable memory, training, and policy for AI agents.
Every session makes the next one better.

Website · Docs · Issues

---

Install

npm install -g @10et/cli

Both tenet and jfl commands are available (backward compatible).

Quick Start

# Add to your project
cd your-project
tenet init

# Set up improvement agents
tenet setup

# Run overnight improvement loop
tenet peter agent test-coverage --rounds 5

# Check results
tenet status

What It Does

TENET is a CLI that gives AI agents persistent memory and a learning loop. Install it alongside Claude Code, Cursor, or any agent.

• Memory — Journals, decisions, and code patterns persist across sessions. Agents don't start from zero.
• Training Loop — Every action produces (state, action, reward) tuples. A policy head learns what works in YOUR codebase.
• Overnight Agents — RL agents try improvements while you sleep. Eval against your metrics, keep what works, revert what doesn't.
• Context Hub — MCP server that any agent connects to. Semantic search across your entire project.
• World Model — Tracks state transitions, detects when assumptions break, predicts outcomes.
• Agent Mesh — P2P network (Subway) for agent-to-agent coordination across machines and sessions.

How It Works

Week 1:  You work normally. TENET captures decisions, outcomes, patterns.
Week 2:  The world model forms. Agents remember what was tried.
Month 1: Agents improve overnight. PRs that make sense for your project.
Month 3: Compound intelligence. New team members inherit full context.

Works With

Any agent runtime — Claude Code, Cursor, Pi, OpenClaw. TENET is the layer underneath.

Migrating from JFL

If you have jfl installed:

npm install -g @10et/cli    # install new package
tenet migrate           # per-project: .jfl/ → .tenet/

The jfl command and .jfl/ directory continue to work — no rush to migrate.

License

MIT