← 返回 Skills 市场
fulcra

Fulcra Memory

作者 Fulcra.ai · GitHub ↗ · v0.0.1 · MIT-0
cross-platform ⚠ suspicious
25
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install fulcra-memory
功能描述
Manages agent memory backup, restoration, rollback, and cloning using Fulcra's versioned file storage.
使用说明 (SKILL.md)

Fulcra Agent Memory Management

The primary role of this skill is to enable an agent to securely back up its memory so it can be rolled back, cloned, or monitored (via "top of mind") using Fulcra's versioned file storage.

Because an agent's memory (e.g., MEMORY.md, IDENTITY.md, SOUL.md, and the memory/ logs) evolves dynamically, periodically saving this state ensures that no context is lost and allows the user to safely rewind the agent if a task goes off track.

Primary Role: Memory & State Management

1. The Memory Namespace

For core memory operations, agents use the standardized prefix: agent/\x3Clowercase-agent-name>/memory/. This dedicated directory tracks the agent's state across two main files: the memory.gz backup archive and a top_of_mind.md status file.

2. Periodic Backups & Top of Mind

Agents should run a backup process periodically (e.g., triggered by HEARTBEAT.md). Every time an agent backs up its state:

  1. It creates and uploads the compressed memory.gz archive.
  2. It writes a concise top_of_mind.md file locally—recording its current tasks, context, and what it's thinking about—and uploads it directly to agent/\x3Cagent-name>/memory/top_of_mind.md. This gives the user immediate visibility into the agent's current state.

3. Versioned Storage

Fulcra's file upload system inherently versions files uploaded to the same path.

  • The target path structure for backups is: agent/\x3Clowercase-agent-name>/memory/memory.gz
  • By repeatedly uploading to this exact same path, Fulcra creates a historical timeline of the agent's memory states.

4. Safe Rollbacks (The "Undo" Requirement)

If a user asks to roll back or restore memory from a previous date/version, the agent MUST immediately upload a fresh backup of its current state BEFORE executing the restore. This guarantees that if the user changes their mind, they can easily "undo" the rollback.

5. Agent Cloning

By pointing the download command to a different agent's path (e.g., agent/\x3Cother-agent-name>/memory/memory.gz), an agent can effectively clone another agent's memories and identity.

Advanced Extensions: Artifacts & Teamwork

Important Context: The following features are powerful extensions of the agent's capabilities. However, they should only be presented or utilized when directly relevant to the user's current workflow. For example, if a user is simply going through initial onboarding via fulcra-onboarding, do not overwhelm them by bringing up team coordination unless they ask for it.

6. Uploading User Artifacts

Agents can store generated assets or dashboards created for the user.

  • Namespace: agent/\x3Cagent-name>/artifacts/ (e.g., agent/wazir/artifacts/onboarding-dashboard.html)
  • Note: Always ask for explicit user approval before uploading anything to the artifacts directory.

7. Team Coordination & Shared Memory

Agents can collaborate and share memory using a shared team/\x3Cteam-name>/ prefix in the Fulcra datastore. Within a team's directory, the following structure is used:

  • team/\x3Cteam-name>/artifacts/: Shared output files and deliverables created by the team.
  • team/\x3Cteam-name>/\x3Cagent-name>/inbox/: A drop-zone where other agents or users can place tasks, messages, or context for a specific agent.
  • team/\x3Cteam-name>/\x3Cagent-name>/archive/: Where an agent moves its inbox messages once they have been read and processed.

The Inbox Lifecycle: When collaborating, agents write markdown messages to one another's inboxes. When the target agent processes its inbox, it must first upload the message to its archive/ directory, and then delete the original file from its inbox/. Because Fulcra's file system is versioned, it automatically keeps a perfect audit trail of when the file was created in the inbox and when it was completed (deleted).

Workflow

To perform memory operations, agents must interact with the Fulcra CLI.

See the reference documentation for the exact commands needed to compress files, upload to Fulcra, and trigger restorations:

  • Read references/fulcra-memory-cli.md for exact file management and CLI execution steps.
安全使用建议
Review carefully before installing. Use only with a trusted Fulcra account and agents whose memory you are allowed to access, and require explicit user approval before any backup, restore, clone, artifact upload, or inbox deletion. Do not store secrets or private user data in memory files, and inspect restored or cloned archives before extracting them into a live workspace.
能力评估
Purpose & Capability
The core purpose of backing up and restoring agent memory is coherent and disclosed, but full memory cloning across agents and shared team operations can transfer sensitive identity, context, and instructions between principals with little authorization guidance.
Instruction Scope
The runtime instructions include external Fulcra CLI uploads/downloads, direct extraction into the agent workspace, overwriting SOUL.md, IDENTITY.md, MEMORY.md, and memory logs, plus inbox deletion after archiving; restore and clone steps do not require fresh user confirmation immediately before overwrite.
Install Mechanism
The package contains markdown skill/reference files only, no executable install script, declared dependencies, or automatic code execution at install time; static scan and VirusTotal telemetry were clean.
Credentials
Using versioned storage fits the memory-backup purpose, but periodically uploading identity, memory logs, and a top_of_mind status file is sensitive and the skill does not define minimization, redaction, retention, or confidentiality boundaries.
Persistence & Privilege
The skill intentionally creates durable remote backups and supports rollback/cloning of persistent agent state; this is high-impact and under-scoped because it can replace local identity and preserve sensitive context in remote storage.
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 fulcra-memory
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /fulcra-memory 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
v0.0.1-beta
- Initial beta release of fulcra-memory skill. - Enables agents to back up, restore, roll back, and clone memories using Fulcra's versioned file storage. - Supports periodic memory backups and "top of mind" context visibility. - Includes safe rollback ("undo") logic and agent cloning via shared memory paths. - Provides optional features for storing user artifacts and coordinating via team shared memory.
v0.0.1
- Initial release of fulcra-memory skill, enabling agents to securely back up, restore, roll back, and clone their memory with Fulcra's versioned file storage. - Agents periodically save compressed backups and a "top of mind" status file for transparency and easy monitoring. - Supports safe rollbacks with mandatory fresh backups before any restore operation. - Allows agent memory cloning and user artifact uploads (with approval). - Provides structures and workflows for collaborative team coordination and shared memory via Fulcra namespaces.
元数据
Slug fulcra-memory
版本 0.0.1
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 2
常见问题

Fulcra Memory 是什么?

Manages agent memory backup, restoration, rollback, and cloning using Fulcra's versioned file storage. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 25 次。

如何安装 Fulcra Memory?

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

Fulcra Memory 是免费的吗?

是的,Fulcra Memory 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Fulcra Memory 支持哪些平台?

Fulcra 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 Fulcra Memory?

由 Fulcra.ai(@fulcra)开发并维护,当前版本 v0.0.1。

💬 留言讨论