ClawHub

Clawhub Memory Tiers

v1.0.0

Tiered memory system for OpenClaw agents. Gives agents instant context on startup with a 4-line state snapshot (L0) and 7-day rolling context (L1). Eliminate...

⭐ 0· 107·0 current·0 all-time
bySwarm Forge@dirtyrootsstudio
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
✓
Purpose & Capability
The name and description claim a local, tiered memory system; the SKILL.md only asks the user/agent to create and read/update L0.md, L1.md, and to add guidance to the agent's SOUL.md. There are no unrelated environment variables, binaries, or installs required.
ℹ
Instruction Scope
The instructions explicitly tell agents to read L0.md and L1.md at startup and to update them at end-of-run — this is within the memory feature's scope. It does require agents to modify their SOUL.md and to read/write files in the agent workspace (including memory/). That is expected for a file-backed memory system but does grant the skill a persistent on-disk footprint in the agent workspace, so operators should be aware of what they choose to store in those files.
✓
Install Mechanism
No install spec, no code files, and no external downloads — instruction-only. Lowest-risk installation surface.
✓
Credentials
No environment variables, credentials, or config paths are requested. The SKILL.md references only files in the agent workspace and the agent's own SOUL.md, which is proportionate to a local memory helper.
✓
Persistence & Privilege
The skill is not always-enabled and does not request elevated platform privileges. It asks agents to persist two small files in their workspace (expected for this feature) and to modify the agent's own SOUL.md; it does not modify other skills or system-wide settings.
Assessment
This skill is an instruction-only pattern for keeping two small text files (L0.md and L1.md) to warm-start agents. It appears coherent and low-risk, but remember: anything you write into these memory files will be read by the agent on every run. Avoid storing secrets, credentials, or sensitive PII in L0/L1. If you run multiple agents, keep each agent's memory files isolated. If you need stricter controls, consider auditing or encrypting what is written to workspace files before adopting the pattern.

Like a lobster shell, security has layers — review code before you run it.

latestvk971vt757tng1dgs5fs1gpkkn1833dfm

License

MIT-0
Free to use, modify, and redistribute. No attribution required.
Termshttps://spdx.org/licenses/MIT-0.html

Comments