Memory Work
ReviewAudited by ClawScan on May 10, 2026.
Overview
This is a transparent local memory/knowledge-management template, but it asks the agent to read and update personal memory files and local notes.
This skill appears purpose-aligned and instruction-only, with no provided code or install script. Use it only in a vault where you are comfortable letting an agent read and organize memory/profile files, review memory writes carefully, and separately vet any executable skills or scripts you add later.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent may read and update files that describe who you are, your preferences, and your long-term memory.
The skill explicitly asks for access to long-term memory, user profile, and identity/collaboration files. This is expected for a memory system, but it is sensitive authority.
这个技能需要访问以下核心文件: `MEMORY.md` - OpenClaw 长期记忆; `USER.md` - 用户信息; `SOUL.md` - 身份文件
Install only in the intended knowledge vault and review proposed changes to USER.md, SOUL.md, and MEMORY.md before accepting them.
Information written into the memory system can be reused in later sessions and may shape the agent’s future recommendations.
The artifacts define persistent memory and startup retrieval that can affect future agent behavior. The confirmation rule helps, but persistent memory still deserves user attention.
On session startup, AI reads only the most recent entry ... Baseline Parameters ... Memory write confirmation: always explicit
Periodically audit MEMORY.md and MEMORY_LOG.md, correct inaccurate memories, and reject entries that should not be retained.
The agent may update weekly task lists, progress notes, and output tables without asking each time.
The skill instructs the agent to automatically scan the Focus Zone and update the weekly file during certain periods. This is aligned with the workflow, but it is automatic local file mutation.
Deep Retrospective (Thursday–Sunday) Auto-execute this; no user trigger required ... Update `_this_week.md`
Keep backups or version control for the vault, and review automatic weekly updates before relying on them.
The current artifact set does not provide executable code, but future or external skill modules could run commands if added to this system.
The documentation describes executable skills and automatic skill invocation, but the supplied package has no code files or install spec. Any separately added scripts would need their own review.
Skills are executable capability modules ... Run scripts or commands ... Execute datetime-check skill ... mandatory first-load every session
Review any added skill folders, scripts, dependencies, and automatic triggers before enabling them.