Skip to content
Back to search

memx

SkillWorks with Claude CodeGitHub

memX: local-first semantic memory for coding agents. Native Claude Code lifecycle hooks.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

Its scripts were only partly scanned.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/NeoLi00/memX with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/NeoLi00/memX'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/neoli00--memx/badge.svg)](/listings/neoli00--memx)

HTML

<a href="/listings/neoli00--memx"><img src="/api/listings/neoli00--memx/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Requirements: Node.js 22.14+ or Node 24. OpenClaw installs require OpenClaw 2026.3.25+. Python 3 is needed only for the default local embedding runtime.

The README commands use the GitHub package spec. A fresh run pulls current GitHub code, so installs do not wait for an npm publish. To use the npm release channel later, replace `github:NeoLi00/memX` with `@neoli00/memx`.

Fill in these values before running a command:

- `--llm-provider`: the provider adapter memX should call. Choose one of `openai-compatible`, `anthropic`, `google`, or `ollama`. - `--llm-base-url`: the base URL for that provider. Examples: `https://api.openai.com/v1`, `https://api.anthropic.com/v1`, `https://generativelanguage.googleapis.com/v1beta`, or `http://127.0.0.1:11434` for Ollama. - `--llm-model`: the model memX uses for memory compilation, recall planning, and maintenance. Pick a fast, low-cost model with reliable JSON output. - `--llm-api-key`: the API key for the provider. Use `--llm-api-key-env PROVIDER_API_KEY` if you want the config to reference an environment variable instead of storing plaintext. For local Ollama, omit the key.

The default embedding setup is local `sentence-transformers-local` with `intfloat/multilingual-e5-small`. Add `--embedding-provider` and `--embedding-model` only when you want to override that default. Use `--dry-run` to preview the files and exec-form commands before writing anything.

For Codex and Claude Code, native hooks are the default lifecycle path for automatic recall and turn capture. Quickstart installs the native plugin, writes the shared memX config, starts or refreshes the managed local memX service, and keeps MCP memory tools hidden with `--mcp-tools none` by default. This prevents duplicate recall/write and prevents the agent from reading audit data as a side channel. Use `--mcp-tools full` only when you intentionally want the agent to see the complete MCP tool set. Generic MCP quickstart stays `full` by default because it has no native lifecycle hooks. Default native memories are also host-scoped, so Codex and Claude Code do not share the same local database unless you deliberately override the database path and actor settings.

If `http://127.0.0.1:3878` is already used by an unmanaged memX-compatible service, quickstart stops instead of silently reusing it. Stop the old service or pass a free local URL, for example `--memx-url http://127.0.0.1:3888`.

Claude Code

This installs the shared memX config, a local Claude Code plugin marketplace, native lifecycle hooks, and the managed local memX service in one run.

npx -y -p github:NeoLi00/memX memx quickstart claude-code \
  --llm-provider openai-compatible \
  --llm-base-url https://llm.example.com/v1 \
  --llm-model fast-memory-model \
  --llm-api-key sk-your-provider-key

Codex

This installs the shared memX config, a local Codex plugin marketplace, native lifecycle hooks, and the managed local memX service in one run.

npx -y -p github:NeoLi00/memX memx quickstart codex \
  --llm-provider openai-compatible \
  --llm-base-url https://llm.example.com/v1 \
  --llm-model fast-memory-model \
  --llm-api-key sk-your-provider-key

OpenClaw

npx -y -p github:NeoLi00/memX memx quickstart openclaw \
  --llm-provider openai-compatible \
  --llm-base-url https://llm.example.com/v1 \
  --llm-model fast-memory-model \
  --llm-api-key sk-your-provider-key

Generic MCP

Use this path for MCP clients that do not have a native memX lifecycle adapter. Quickstart writes the shared memX config, starts the managed local memX service, and prints a ready-to-copy MCP server config.

npx -y -p github:NeoLi00/memX memx quickstart mcp \
  --llm-provider openai-compatible \
  --llm-base-url https://llm.example.com/v1 \
  --llm-model fast-memory-model \
  --llm-api-key sk-your-provider-key

Service management

npx -y -p github:NeoLi00/memX memx service status

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/plugin.json · Manifest · 11 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. {
  2. "name": "memx",
  3. "version": "2026.3.15",
  4. "description": "memX: local-first semantic memory for coding agents. Native Claude Code lifecycle hooks.",
  5. "author": {
  6. "name": "Neo Li"
  7. },
  8. "license": "MIT",
  9. "homepage": "https://github.com/NeoLi00/memX",
  10. "repository": "https://github.com/NeoLi00/memX"
  11. }
README.md · README · 198 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. <p align="center">
  2. <img src="./assets/memx-cover-en.svg" alt="memX - self-learning, self-maintaining memory for AI agents" width="920">
  3. </p>
  4. <p align="center">
  5. <a href="./README.md">English</a> · <a href="./README-ch.md">中文</a> ·
  6. <a href="./ARCHITECTURE.md">Architecture</a>
  7. </p>
  8. ---
  9. memX turns completed work into structured, searchable, self-maintained memory, then injects only the evidence an agent needs for the current query.
  10. It connects natively to Codex, Claude Code, and OpenClaw, and reaches any MCP-compatible client through the same local memory layer.
  11. ## Benchmarks
  12. <table align="center">
  13. <thead>
  14. <tr>
  15. <th>Suite</th>
  16. <th>Scope</th>
  17. <th>R@3 success rate</th>
  18. </tr>
  19. </thead>
  20. <tbody>
  21. <tr>
  22. <td><strong>LongMemEval-S</strong></td>
  23. <td>Long-context memory retrieval</td>
  24. <td><strong>94.2%</strong></td>
  25. </tr>
  26. <tr>
  27. <td><strong>Real engineering cases</strong></td>
  28. <td>30 cases, each with 20+ turns</td>
  29. <td><strong>100%</strong></td>
  30. </tr>
  31. </tbody>
  32. </table>
  33. ## Architecture
  34. <p align="center">
  35. <img src="./assets/memx-overview.svg" alt="memX coarse architecture" width="920">
  36. </p>
  37. ## Agent support
  38. <table align="center">
  39. <tr>
  40. <td align="center" width="56"><img src="./assets/agent-logos/codex.png" alt="Codex logo" width="34"></td>
  41. <td><strong>Codex</strong></td>
  42. <td><sub>native hooks, MCP hidden by default</sub></td>
  43. </tr>
  44. <tr>
  45. <td align="center" width="56"><img src="./assets/agent-logos/claude-code.png" alt="Claude Code logo" width="34"></td>
  46. <td><strong>Claude Code</strong></td>
  47. <td><sub>native hooks, MCP hidden by default</sub></td>
  48. </tr>
  49. <tr>
  50. <td align="center" width="56"><img src="./assets/agent-logos/openclaw.png" alt="OpenClaw logo" width="34"></td>
  51. <td><strong>OpenClaw</strong></td>
  52. <td><sub>native + hooks</sub></td>
  53. </tr>
  54. <tr>
  55. <td align="center" width="56"><strong>MCP</strong></td>
  56. <td><strong>MCP clients</strong></td>
  57. <td><sub>any MCP-compatible client</sub></td>
  58. </tr>
  59. </table>
  60. ## Quick start
  61. Requirements: Node.js 22.14+ or Node 24. OpenClaw installs require OpenClaw 2026.3.25+. Python 3 is
  62. needed only for the default local embedding runtime.
  63. The README commands use the GitHub package spec. A fresh run pulls current GitHub code, so installs
  64. do not wait for an npm publish. To use the npm release channel later, replace
  65. `github:NeoLi00/memX` with `@neoli00/memx`.
  66. Fill in these values before running a command:
  67. - `--llm-provider`: the provider adapter memX should call. Choose one of `openai-compatible`,
  68. `anthropic`, `google`, or `ollama`.
  69. - `--llm-base-url`: the base URL for that provider. Examples: `https://api.openai.com/v1`,
  70. `https://api.anthropic.com/v1`, `https://generativelanguage.googleapis.com/v1beta`, or
  71. `http://127.0.0.1:11434` for Ollama.
  72. - `--llm-model`: the model memX uses for memory compilation, recall planning, and maintenance.
  73. Pick a fast, low-cost model with reliable JSON output.
  74. - `--llm-api-key`: the API key for the provider. Use `--llm-api-key-env PROVIDER_API_KEY` if you
  75. want the config to reference an environment variable instead of storing plaintext. For local
  76. Ollama, omit the key.
  77. The default embedding setup is local `sentence-transformers-local` with
  78. `intfloat/multilingual-e5-small`. Add `--embedding-provider` and `--embedding-model` only when you
  79. want to override that default. Use `--dry-run` to preview the files and exec-form commands before
  80. writing anything.
  81. For Codex and Claude Code, native hooks are the default lifecycle path for automatic recall and
  82. turn capture. Quickstart installs the native plugin, writes the shared memX config, starts or
  83. refreshes the managed local memX service, and keeps MCP memory tools hidden with `--mcp-tools none`
  84. by default. This prevents duplicate recall/write and prevents the agent from reading audit data as a
  85. side channel. Use `--mcp-tools full` only when you intentionally want the agent to see the complete
  86. MCP tool set. Generic MCP quickstart stays `full` by default because it has no native lifecycle
  87. hooks. Default native memories are also host-scoped, so Codex and Claude Code do not share the same
  88. local database unless you deliberately override the database path and actor settings.
  89. If `http://127.0.0.1:3878` is already used by an unmanaged memX-compatible service, quickstart stops
  90. instead of silently reusing it. Stop the old service or pass a free local URL, for example
  91. `--memx-url http://127.0.0.1:3888`.
  92. ### Claude Code
  93. This installs the shared memX config, a local Claude Code plugin marketplace, native lifecycle
  94. hooks, and the managed local memX service in one run.
  95. ```bash
  96. npx -y -p github:NeoLi00/memX memx quickstart claude-code \
  97. --llm-provider openai-compatible \
  98. --llm-base-url https://llm.example.com/v1 \
  99. --llm-model fast-memory-model \
  100. --llm-api-key sk-your-provider-key
  101. ```
  102. ### Codex
  103. This installs the shared memX config, a local Codex plugin marketplace, native lifecycle hooks, and
  104. the managed local memX service in one run.
  105. ```bash
  106. npx -y -p github:NeoLi00/memX memx quickstart codex \
  107. --llm-provider openai-compatible \
  108. --llm-base-url https://llm.example.com/v1 \
  109. --llm-model fast-memory-model \
  110. --llm-api-key sk-your-provider-key
  111. ```
  112. ### OpenClaw
  113. ```bash
  114. npx -y -p github:NeoLi00/memX memx quickstart openclaw \
  115. --llm-provider openai-compatible \
  116. --llm-base-url https://llm.example.com/v1 \
  117. --llm-model fast-memory-model \
  118. --llm-api-key sk-your-provider-key
  119. ```
  120. ### Generic MCP
  121. Use this path for MCP clients that do not have a native memX lifecycle adapter. Quickstart writes the
  122. shared memX config, starts the managed local memX service, and prints a ready-to-copy MCP server
  123. config.
  124. ```bash
  125. npx -y -p github:NeoLi00/memX memx quickstart mcp \
  126. --llm-provider openai-compatible \
  127. --llm-base-url https://llm.example.com/v1 \
  128. --llm-model fast-memory-model \
  129. --llm-api-key sk-your-provider-key
  130. ```
  131. ### Service management
  132. ```bash
  133. npx -y -p github:NeoLi00/memX memx service status
  134. npx -y -p github:NeoLi00/memX memx service restart
  135. npx -y -p github:NeoLi00/memX memx service stop
  136. ```
  137. Use the same `--home`, `--memx-url`, and `--memx-secret` values that you used during quickstart when
  138. you manage a non-default install.
  139. ## Clean uninstall
  140. Each uninstall command backs up the target config first, then removes only memX-owned entries.
  141. Claude Code and Codex cleanup also stop the managed local service, uninstall the native plugin,
  142. remove the local marketplace, and delete the generated marketplace snapshot.
  143. OpenClaw cleanup also removes stale `memx` / `memory-memx` slot, allow, and entry references, then
  144. best-effort uninstalls both current and legacy plugin files if OpenClaw can still see them.
  145. ```bash
  146. npx -y -p github:NeoLi00/memX memx uninstall openclaw
  147. npx -y -p github:NeoLi00/memX memx uninstall codex
  148. npx -y -p github:NeoLi00/memX memx uninstall claude-code
  149. ```
  150. Add `--dry-run` to preview, or `--config /path/to/config` when using a non-default config path.
  151. ## What memX can do
  152. - **Remember work over time**: project decisions, user preferences, task status, long source
  153. segments, and raw evidence stay linked to the original turn.
  154. - **Connect related things**: projects, repos, tools, files, resources, blockers, and outcomes can
  155. be represented as entities and graph edges.
  156. - **Learn collaboration patterns**: repeated evidence can become reusable guidance without losing
  157. its supporting sources.
  158. - **Maintain itself**: corrections can supersede older facts, stable evidence can be promoted, and
  159. stale task state stops competing with current state.
  160. - **Recall compact evidence**: facts, events, state, chunks, relationships, resources, and learned
  161. patterns are searched together, then injected as small evidence lines.

Version history

  • v2026.3.1510/19/2026

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.