← 返回 Skills 市场
leegitw

Memory Garden - Validated Community Memory

作者 Lee Brown · GitHub ↗ · v1.0.0
cross-platform ✓ 安全检测通过
272
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install memory-garden
功能描述
N-count validated knowledge for AI agents. Patterns that prove themselves through repeated use. Local-first, community-ready.
安全使用建议
This skill appears to do what it claims — run a local mg-daemon, augment queries with validated patterns, and optionally extract/sync patterns. Before installing: 1) Verify the release binaries (download URL + SHA256 in SKILL.md) against the upstream GitHub release you trust. 2) Be aware the skill will create ~/.memory-garden and generate an Ed25519 private key (identity.json, mode 0600) used for signing; protect and back up this file or override its location via MG_IDENTITY_DIR if needed. 3) Extraction and P2P sync are opt-in (disabled by default) — do not enable MG_EXTRACTION_ENABLED or MG_SYNC_ENABLED unless you want the agent to extract conversation content or participate in networked syncing. 4) If you prefer the skill not to augment queries automatically, disable search with MG_SEARCH_ENABLED=false. 5) Inspect scripts/install.sh and mg-daemon upstream source if you need to audit what the native daemon binary does when run with --serve. Overall the required permissions and behavior are proportional to the stated purpose.
功能分析
Type: OpenClaw Skill Name: memory-garden Version: 1.0.0 The OpenClaw skill 'memory-garden' is designed for local-first knowledge management and exhibits strong security practices. It manages a local daemon (`mg-daemon`) with robust lifecycle handling, including symlink-protected PID file management and verification that only its own processes are terminated. Cryptographic identity management uses Ed25519 keys stored with strict permissions and includes replay protection for signed messages. Communication with the local daemon is via secure HTTP POST requests with JSON bodies, preventing shell injection. The `SKILL.md` and code do not contain any prompt injection attempts or evidence of intentional malicious behavior like data exfiltration, unauthorized remote control, or persistence mechanisms beyond its stated purpose. All external binary downloads are accompanied by SHA256 hashes, and sensitive operations are carefully handled with input validation and error recovery.
能力评估
Purpose & Capability
Name/description match behavior: the skill runs a local 'mg-daemon', stores patterns under ~/.memory-garden, and provides search/extract/validate tools. The declared required binary (mg-daemon) and the code (daemon manager + MCP tool calls) are appropriate for a local-first memory service.
Instruction Scope
SKILL.md and code explicitly start and manage a local daemon, perform health checks, read/write under ~/.memory-garden, and optionally enable P2P sync. This is within scope, but note: the skill will generate and persist an Ed25519 private key (identity.json) under ~/.memory-garden/identity by default; that private key is sensitive and is used for signing P2P attestations. Also the runtime references additional env vars (MG_IDENTITY_DIR, MG_DATA_DIR) in code/docstrings that are not enumerated in the registry metadata — they are optional but affect where files are stored.
Install Mechanism
No installation spec is required by the registry entry, but SKILL.md metadata includes download entries pointing to GitHub releases for macOS and Linux with SHA256 sums — a well-known release host. Downloads are expected and reasonable for a native daemon binary. No arbitrary paste sites or IP URLs are used.
Credentials
The skill does not request unrelated cloud credentials. It reads configuration via MG_* env vars (MG_DAEMON_URL, MG_EXTRACTION_ENABLED, MG_EXTRACTION_CONFIRM, MG_SYNC_ENABLED, MG_SEARCH_LIMIT) documented in SKILL.md. The code also recognizes MG_IDENTITY_DIR and refers to MG_DATA_DIR in error messaging (these are optional). It generates and stores a private key locally (sensitive) — required for P2P signing, so justified, but the user should treat that artifact as a secret.
Persistence & Privilege
always:false (not force-included) and disable-model-invocation:false (normal). The skill will autonomously run beforeQuery/afterResponse hooks when invoked and may start a detached daemon process and write files to ~/.memory-garden. This level of persistence is expected for a local daemon skill; there is no indication it modifies other skills or system-wide agent settings.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install memory-garden
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /memory-garden 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Initial release of Memory Garden - Introduces N-count validated knowledge management for AI agents. - Local-first design keeps patterns private unless you opt in to share. - Supports macOS (arm64, amd64), Linux (amd64), and WSL2 for Windows. - Features include pattern search, extraction, N-count tracking, and optional P2P sync. - Fully configurable via environment variables; detailed installation and troubleshooting guides provided. - Community-ready with focus on pattern validation, privacy, and trustworthy AI memory.
元数据
Slug memory-garden
版本 1.0.0
许可证
累计安装 1
当前安装数 1
历史版本数 1
常见问题

Memory Garden - Validated Community Memory 是什么?

N-count validated knowledge for AI agents. Patterns that prove themselves through repeated use. Local-first, community-ready. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 272 次。

如何安装 Memory Garden - Validated Community Memory?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install memory-garden」即可一键安装,无需额外配置。

Memory Garden - Validated Community Memory 是免费的吗?

是的,Memory Garden - Validated Community Memory 完全免费(开源免费),可自由下载、安装和使用。

Memory Garden - Validated Community Memory 支持哪些平台?

Memory Garden - Validated Community Memory 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Memory Garden - Validated Community Memory?

由 Lee Brown(@leegitw)开发并维护,当前版本 v1.0.0。

💬 留言讨论