Store and retrieve long-term user context in an append-only JSONL memory file.
npx skills add https://github.com/Hongyu-yu/matsci-ai-skills --skill memory-hongyu-yu
Please help me install this Agent Skill. Skill: memory Source: https://github.com/Hongyu-yu/matsci-ai-skills/tree/main/skills/memory Command: npx skills add https://github.com/Hongyu-yu/matsci-ai-skills --skill memory-hongyu-yu
This Skill provides a persistent, file-based memory for AI prompts, enabling context to be remembered across sessions without re-asking the user.
Configure MEMORY_FILE if needed, then add an entry with memory_write and query it with memory_search.
High-intent search queries and answers about installing and using this skill.
You can keep long-term memory across chat sessions by using an append-only JSONL file to store and retrieve user context. This approach ensures data integrity while allowing the AI to recall past preferences and decisions in ongoing conversations without re-asking the user.
You can append and search structured memory entries using bash and jq by utilizing dedicated write and search interfaces. The write interface appends JSON entries with fields like ts, type, content, tags, and meta, while the search interface queries recent context with safe escaping via jq.
Yes, you need jq installed to use this append-only JSONL memory file approach. jq is a required dependency for ensuring strict JSON formatting and safe escaping when writing to and searching the persistent memory file.
The best way to store persistent user context without a database is using an append-only JSONL memory file. It provides a lightweight, file-based solution for AI agents requiring persistent memory, context recall, and recent-event search across prompts.
Yes, you can retrieve recent events and tasks from long-running AI interactions using the memory search interface. It queries the append-only JSONL memory file with a focus on recency, enabling rapid retrieval of relevant notes, facts, and past decisions.
Memory search focuses on recency in an append-only JSONL file because it prioritizes the latest context for ongoing conversations. This ensures the AI assistant retrieves the most relevant recent notes and user preferences to tailor responses effectively.
aresbitSearch and analyze historical session JSONL files using jq and rg.
MOODMNKY-LLCSearch and analyze historical JSONL session logs with ripgrep and jq.
AcosmiTrigger agent runs from external HTTP webhook events.
arunodaSearch extracted facts locally using Ollama embeddings and sqlite-vec.
