aura-memory-bench

Benchmark Aura's memory extraction, retrieval, and QA pipeline against production-like timelines.

1|4|Updated Feb 13, 2026

Search Tags:#timeline #recall #qa #memory #aura #benchmark #as-of

Ready to run

npx skills add https://github.com/AuraHQ-ai/aura --skill aura-memory-bench

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: aura-memory-bench
Source: https://github.com/AuraHQ-ai/aura/tree/main/.cursor/skills/aura-memory-bench
Command: npx skills add https://github.com/AuraHQ-ai/aura --skill aura-memory-bench

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

End-to-end harness that replays production-like conversations through Aura's memory pipeline to measure extraction, recall, and QA performance, enabling deterministic benchmarking and regression tracking.

Core Features & Use Cases

End-to-end timeline replay: replicates real interactions and evaluations against the production memory system.
Deterministic scoring: computes recall@15 and QA accuracy using a consistent evaluation flow.
Safe benching: isolates bench logic from production memory code to prevent data leakage or side effects.

Quick Start

Run the memory bench from the repository root to execute Aura's memory bench against the production-like timeline.

Frequently Asked Questions about aura-memory-bench

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I benchmark memory recall and temporal reasoning against production timelines?▼

The memory benchmark evaluates extraction, recall, and QA pipeline performance by replaying timelines against the production memory system. It computes deterministic scores like recall@15 and QA accuracy to track regressions safely.

What do I need to run a memory benchmark with timeline replay?▼

You need a live DATABASE_URL and AI Gateway access to execute the memory bench. The harness orchestrates timeline replay using production memory functions to safely evaluate extraction and recall without data leakage.

Can I use this memory benchmark harness for regression tracking in CI?▼

Yes, you can run the memory bench in CI or local environments for deterministic regression tracking. The harness computes recall@15 and QA accuracy against toy and LongMemEval corpora to measure knowledge-update behavior consistently.

How does the memory bench isolate evaluation logic from production memory code?▼

The harness isolates bench logic from production memory code to prevent data leakage or side effects during evaluation. Bench tooling operates separately while orchestrating timeline replay through the production memory functions.

What evaluation metrics does the memory pipeline benchmark compute?▼

The benchmark computes deterministic scoring metrics including recall@15 and QA accuracy. These metrics evaluate end-to-end memory recall, temporal reasoning, and knowledge-update behavior across toy and LongMemEval corpora.