skills.rest
ayvazyan10
ayvazyan10COMMUNITY
Explore all skills by this author →
WebsiteGitHub

engram

Store and recall AI memories across sessions using local embeddings and a knowledge graph.

9|2|Updated Mar 21, 2026
Search Tags:#knowledge graph#semantic search#persistent memory#context injection#contradictions#local embeddings
One-click install
npx skills add https://github.com/ayvazyan10/engram --skill engram-ayvazyan10
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: engram
Source: https://github.com/ayvazyan10/engram/tree/main
Command: npx skills add https://github.com/ayvazyan10/engram --skill engram-ayvazyan10

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Engram solves the problem of AI forgetting important context between sessions by providing a persistent, structured memory backend with automatic retrieval and contradiction awareness.

Core Features & Use Cases

  • Persistent AI memory: store episodic events, semantic facts, and procedural patterns in a local SQLite database.
  • Semantic recall pipeline: retrieve relevant memories using vector similarity, expand results via a knowledge graph, score and rank candidates, then assemble a token-budgeted context block.
  • Contradiction detection & resolution: detect conflicting beliefs on store and resolve them with configurable strategies.
  • Memory lifecycle management: run decay sweeps (Ebbinghaus forgetting) and consolidate older episodic memories into semantic summaries.
  • Operational surfaces: use via MCP tools (Claude Code), REST API (server), CLI (terminal), and adapters (Ollama proxy, OpenClaw integration).

Quick Start

Activate Engram by running a recall at the start of your workflow, using the query you want your AI to remember.

Frequently Asked Questions about engram

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

FAQPage Schema
How do I give my AI persistent memory across sessions?▼

Persistent AI memory stores episodic events, semantic facts, and procedural patterns in a local SQLite database, enabling cross-session recall. Engram provides this backend using local embeddings and a knowledge graph to automatically retrieve relevant context for your AI.

How does semantic recall work with a knowledge graph for context injection?▼

Semantic recall uses vector similarity search over an HNSW index to find relevant memories, expands results via BFS graph traversal, scores candidates by importance and recency, and assembles a token-budgeted context block for prompt injection.

Can I detect and resolve contradictions in AI memory automatically?▼

Contradiction detection identifies conflicting beliefs on store and resolves them using configurable strategies. This prevents your AI from holding inconsistent facts across different sessions when persistent memory is updated.

Does persistent AI memory work with MCP and REST API workflows?▼

Persistent memory integrates via MCP tools for Claude Code, a REST API server, a CLI for terminal workflows, and adapters for Ollama proxy and OpenClaw integration. These surfaces allow flexible context assembly across different development environments.

Do I need Node.js 22 to run local embeddings for AI memory?▼

Node.js 22+ compatibility is required to run local ONNX/WASM embedding computation. This local processing enables vector similarity search without external API dependencies, keeping your AI memory entirely self-contained.

What is the best way to manage memory lifecycle and consolidate old episodic events?▼

Memory lifecycle management runs decay sweeps using Ebbinghaus forgetting curves and consolidates older episodic memories into semantic summaries. This keeps your persistent AI memory optimized and prevents stale context from accumulating.

Related Skills

Explore More Software Engineering Skills→
aarushlohitaarushlohit

chromadb

Store, search, backup, and reset ChromaDB vector database entries via Python scripts.

Community
Advanced
boshu2boshu2
418

inject

Inject prior knowledge from .agents artifacts into session context at SessionStart.

Community
Basic
ollieb89ollieb89

ollama-local

Manage local Ollama models and run chat, completions, and embeddings.

Community
Intermediate
FalkorDBFalkorDB
22

Create and query vector indexes

Create and query HNSW vector indexes in FalkorDB graph databases.

Official
Intermediate