v0.3.0 — open source

Make LLMs respect
your decisions

Stop repeating constraints. Stop prompt drift. Mneme injects relevant project decisions only when they matter.

terminal
$ mneme check --mode strict
Checking decisions against current context...

PASS  Storage decision enforced — Postgres locked, no new DBs
PASS  Auth pattern respected — JWT middleware unchanged
WARN  New dependency introduced — prisma not in approved list
FAIL  Violates ADR-004 — Repository pattern bypassed in user.service.ts

2 passed · 1 warning · 1 failure

Watch a 17-second demo

Mneme catching architectural drift before it reaches your AI coding agent.

Watch Demo →

Decision enforcement for AI coding workflows

🗃️

Capture decisions

Store decisions in structured files that travel with your codebase — version controlled, repo-native, always in sync.

🎯

Selective injection

Retrieve only what matters per task. No bloated prompts — just the decisions relevant to what you're building now.

🛡️

Enforce consistency

Run mneme check before committing or in CI. Get PASS / WARN / FAIL signals before the model drifts.

Cursor + Claude Code

Generate Cursor rules from your decisions with mneme cursor generate. Zero config, instant enforcement.

Inject, don't rely on memory

1

Add a decision

Run mneme add "Use Postgres — no new DBs". It's stored in your project's decision file and committed with your code.

2

Check before you ship

Run mneme check --mode strict in your terminal or CI pipeline. It validates your current codebase against every stored decision.

3

Generate rules for your editor

Run mneme cursor generate to output .cursor/rules/mneme.mdc — your decisions become editor-level guardrails automatically.

4

LLMs stay aligned

Every AI suggestion in your workflow gets filtered through your decisions. No more explaining the same constraints session after session.