agentsclimarketplace

Nano memory

Skill quynj/quynj-claw/.agents/skills/nano-memory

本地 Agent 会话控制台 — 基于 AgentScope Java 的轻量级 AI 对话运行时

Install
npx -y skills add quynj/quynj-claw --skill nano-memory

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Maintain durable local memory for the Quynj Claw user and project using only the registered file tools and shell search. Use when the agent needs to recall who the user is, user preferences, project history, prior decisions, daily logs, or when it learns durable facts that should survive new sessions.

SKILL.md

3.1 KB, as published. Nobody here has run it

Nano Memory

Architecture

Durable memory lives in project-local files. AgentScope session memory and UI JSON files are not a substitute for this skill.

Core files:

  • .agents/memory/MEMORY.md: curated long-term memory for user identity, preferences, project conventions, and stable decisions.
  • .agents/memory/YYYY-MM-DD.md: daily chronological logs for recent work and raw notes.

Available Tools

Use only tools that are actually registered in this Quynj Claw runtime:

  • list_files or list_directory to inspect directories.
  • view_text_file to read file contents, optionally with line ranges.
  • write_text_file to create, overwrite, or replace a range in a text file.
  • insert_text_file to insert text at a specific line when available.
  • Shell search commands such as rg, grep, ls, find, or platform equivalents when the runtime allows shell access.

Do not call non-existent tools such as read_file, write_file, or edit_file unless the current runtime explicitly exposes them. In this project, prefer view_text_file, write_text_file, and insert_text_file.

Workflow A: Retrieve Memory

  1. List .agents/memory/ to see available memory files.
  2. Search .agents/memory/ for relevant names, topics, dates, decisions, or preferences.
  3. Read the specific matching files with view_text_file.
  4. Answer based on retrieved memory. If memory is absent or ambiguous, say so.

Use this workflow before claiming to know who the user is, what they prefer, or what happened in previous sessions.

Workflow B: Record Daily Context

  1. Use the current local date for .agents/memory/YYYY-MM-DD.md.
  2. If the file exists, read it first with view_text_file; then append by rewriting the file with write_text_file or insert at the end with insert_text_file.
  3. If the file does not exist, create it with write_text_file.
  4. Keep daily entries short, dated, and factual.

Workflow C: Update Long-Term Memory

  1. Read .agents/memory/MEMORY.md with view_text_file before updating.
  2. Preserve existing structure and useful content.
  3. Use write_text_file with a precise line range when the target section is clear. Otherwise rewrite the full file after reading the latest contents.
  4. Record durable facts only: stable preferences, identity details the user wants remembered, project conventions, architecture decisions, recurring workflows, and completed milestones.
  5. Do not store passwords, API keys, private medical/financial details, or sensitive personal data unless the user explicitly requests it.

Recording Rules

  • Record first, then rely on the memory in future answers.
  • Prefer concise distilled memory over full chat transcripts.
  • Do not use .agents/ui-store as memory; those files are UI projections only.
  • If a write tool fails, report the failure and do not pretend the memory was saved.

Gives 0 of the 12 instructions most memory context skills give

Counted across 674 of the 847 authors here whose files we hold, read 2026-08-06

  • inform the user when setup is completein 21 of 674, across 6 files
  • confirm the draft with the user before writingin 21 of 674, across 6 files
  • update the agent skills block in place if it existsin 21 of 674, across 6 files
  • present findings to the userin 20 of 674, across 5 files
  • write the three docs files from seed templatesin 20 of 674, across 5 files
  • ask the user about each decision one at a timein 19 of 674, across 4 files
  • edit CLAUDE.md if it existsin 18 of 674, across 3 files
  • explore current repo statein 18 of 674, across 3 files
  • do not overwrite user edits to surrounding sectionsin 18 of 674, across 3 files
  • back up the original file before overwritingin 16 of 674, across 8 files
  • keep the memory index under 200 linesin 15 of 674
  • Provide actionable steps and verificationin 13 of 674, across 2 files

Said here and by no other author read

  • store durable local memory in project files
  • use only registered file and shell tools
  • report absent or ambiguous memory honestly
  • create dated daily logs for recent work
  • preserve existing structure when updating memory
  • store concise distilled memory over full transcripts

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.