enigma-memory v0.1.4

Benchmark optimization release.

Install:

npm install -g enigma-memory
enigma quickstart --bundle ./.enigma/bundle.json --overwrite
enigma demo cross-model
npm run benchmark:memory-suite

Highlights:

• adds query-aware relevance filtering before context-pack optimizer tiering
• improves deterministic local benchmark Enigma context-pack footprint from prior 281 total estimated tokens / 56.2 mean tokens to 148 total / 29.6 mean while keeping exact recall and abstention at 1.0 on the fixture
• keeps external competitor rows requirements-only for Letta/MemGPT, LangGraph, Zep, Mem0, OpenAI native memory, and Claude memory tool until credentials/runtimes/datasets are supplied

Verification:

• node --test test/enigma-memory-benchmarks.test.mjs test/enigma.test.mjs
• npm run benchmark:memory-suite
• npm run check
• npm test
• npm pack --dry-run
• npm publish --access public --dry-run
• npm publish --access public
• npm run registry:verify -- --execute
• registry install smoke for benchmark/quickstart/cross-model
• GitHub CI passed on Ubuntu and Windows

Boundaries:
No provider deletion proof, model forgetting proof, ROI/savings guarantee, compliance certification, third-party competitor score, or benchmark-leadership claim.