Your coding agent has amnesia. Here's the memory architecture that fixes it.

Описание и применение

Архитектура Engraphis обеспечивает долговременную, ограниченную по контексту и проверяемую память для кода, позволяя агенту хранить решения, связанные с репозиторием, и быстро получать их без повторного ввода.

Задача

Coding agents lose durable project knowledge when context windows are compacted or sessions end, causing them to forget prior decisions and propose incorrect actions.

Результат

Engraphis reduces retrieved context tokens by 71.1% on deterministic fixtures, provides durable, scoped, provenanced memory, and enables agents to cite decisions and avoid repeating past mistakes.

Рабочий процесс

• 1. Install Engraphis MCP or dashboard.
• 2. Initialize memory service with a local SQLite database.
• 3. Use engraphis_remember to store decisions with workspace, repo, and provenance.
• 4. Use engraphis_recall_context to retrieve compact, scoped context for the current task.
• 5. Agent uses retrieved context to inform decisions and can cite provenance or abstain if uncertain.

Доказательства

A coding agent has three kinds of state: the current task, the working context, and durable project knowledge that must survive the session.

Источник

Engraphis is deliberately boring at the storage layer and careful at the retrieval layer.

Источник

The result is the same: the agent has information, but not context with provenance.

Источник

Mean retrieved context dropped from 740.3 to 214.3 tokens per question — a 71.1% reduction across 18 questions.

Источник