← 返回 Skills 市场
1895
总下载
1
收藏
2
当前安装
2
版本数
在 OpenClaw 中安装
/install shodh-local
功能描述
Local offline cognitive memory for AI agents providing semantic recall, GTD todos/projects, and a knowledge graph with Hebbian learning and a TUI dashboard.
使用说明 (SKILL.md)
Shodh-Local (v0.1.74)
Local-first brain for OpenClaw. Offline, learns with use.
Config (TOOLS.md)
- Binary:
./shodh-memory-server(or add to PATH) - Server:
localhost:3030 - Data:
./shodh-data - Key:
\x3CYOUR-API-KEY>(X-API-Key, generate via shodh-memory-server) - Manage:
processtool (sessionamber-seaslug) - TUI:
cd tools/shodh-memory && ./shodh-tui(graph/activity)
Quick Use
KEY='\x3CYOUR-API-KEY>'
curl -s -X POST http://localhost:3030/api/remember \\
-H "Content-Type: application/json" -H "X-API-Key: $KEY" \\
-d '{"user_id": "henry", "content": "Test memory", "memory_type": "Learning", "tags": ["test"]}'
Core Tools
- Remember:
/api/remember(types: Learning/Observation/Conversation/Task/Preference) - Recall:
/api/recall(semantic) |/api/recall/tags - Proactive:
/api/proactive_context(auto-relevant) - Todos:
/api/todos/add|/api/todos|/api/todos/complete - Projects:
/api/projects/add|/api/projects - Summary:
/api/context_summary
Full API: reference/api.md
Best Practices
- User ID:
henry(main),openclaw(system),task-XYZ(sub-agents) - Tags: Always add for filtering (e.g. ["openclaw", "project-backend"])
- Before reply: Recall recent context for continuity
- Heartbeat: Check todos daily
- Maintenance: Restart server weekly (
process kill amber-seaslug+ restart)
Read reference/examples.md for OpenClaw patterns.
安全使用建议
This skill appears to be a local memory/server integration, but the published metadata omits required pieces the SKILL.md expects. Before installing: (1) confirm where ./shodh-memory-server comes from and that you trust that binary (inspect its origin or source code); (2) ensure you understand and control the suggested X-API-Key (keep it local and do not reuse other service credentials); (3) ask the skill author to update the manifest to declare required binaries, environment variables, and any tools (e.g., the 'process' tool) so you can review privileges; (4) verify the agent will only talk to localhost:3030 and not a remote host or reverse proxy; and (5) avoid running any untrusted binaries or giving the agent system-level process control until the above are clarified.
功能分析
Type: OpenClaw Skill
Name: shodh-local
Version: 1.0.1
The skill bundle describes a 'local-first cognitive memory' system that runs a local server (`./shodh-memory-server`) and interacts with it via `localhost:3030`. The `SKILL.md` instructs the OpenClaw agent to manage this local process using a `process` tool and to use `curl` commands to interact with the local API, authenticated by an API key. While running local executables and managing processes are powerful capabilities, they are explicitly stated as core to the skill's purpose. There is no evidence of data exfiltration to external endpoints, unauthorized access, persistence mechanisms, or deceptive prompt injection attempts against the agent. All interactions are confined to the local machine and the skill's stated functionality.
能力评估
Purpose & Capability
The SKILL.md clearly describes a local server binary (./shodh-memory-server), a data directory (./shodh-data), an X-API-Key, and a process management session (amber-seaslug). However the registry metadata lists no required binaries, env vars, or config paths. That inconsistency means the declared manifest does not match the actual runtime expectations.
Instruction Scope
Runtime instructions are focused on running and interacting with a local memory server (endpoints on localhost:3030) and are consistent with a local-first memory skill. They do ask the agent to manage the server process (e.g., `process kill amber-seaslug` + restart) and to 'recall recent context before reply'. Those process-management steps reference a 'process' tool/session that isn't described in the metadata, which expands the agent's actions beyond simple HTTP calls but does not by itself indicate exfiltration or remote endpoints.
Install Mechanism
This is an instruction-only skill with no install spec and no code files — lowest install risk. Nothing in the manifest attempts to download or install remote code. The SKILL.md assumes a preinstalled binary, which is fine, but that binary is not declared in the metadata.
Credentials
The examples and TOOLS.md reference an API key (X-API-Key) and show usage via a KEY environment variable, but the skill declares no required environment variables and no primary credential. The single credential (local API key) would be proportionate to the stated purpose, but the omission from the manifest is an inconsistency the user should resolve before trusting the skill.
Persistence & Privilege
The skill does not request always:true or other elevated persistence. It does instruct process management (killing/restarting a local session) which may require system-level access depending on how the agent executes tools; the manifest does not declare or justify such privileges. Autonomous invocation is allowed (default) but not on its own a red flag.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install shodh-local - 安装完成后，直接呼叫该 Skill 的名称或使用
/shodh-local触发 - 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
v1.0.1
- Added a summary field for improved discoverability.
- Updated the binary path in config instructions for clarity (now suggests ./shodh-memory-server or adding to PATH).
v1.0.0
- Initial release of shodh-local: local Shodh-Memory v0.1.74 for persistent, offline cognitive memory.
- Supports remembering, semantic recall, GTD-style todos, projects, and knowledge graph operations.
- Offers RESTful API endpoints for memory, recall (semantic and by tag), proactive context, todos, and projects.
- Features 3-tier Hebbian learning memory (working/session/LTM), a TUI dashboard, and best practices for agent use.
- Quick setup and API usage instructions included for local/offline deployment.
元数据
常见问题
Shodh Local 是什么？
Local offline cognitive memory for AI agents providing semantic recall, GTD todos/projects, and a knowledge graph with Hebbian learning and a TUI dashboard. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 1895 次。
如何安装 Shodh Local？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install shodh-local」即可一键安装，无需额外配置。
Shodh Local 是免费的吗？
是的，Shodh Local 完全免费（开源免费），可自由下载、安装和使用。
Shodh Local 支持哪些平台？
Shodh Local 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 Shodh Local？
由 doobidoo（@doobidoo）开发并维护，当前版本 v1.0.1。
推荐 Skills