Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Agent Memory Setup
v1.2.0Set up the full OpenClaw agent memory system with 3-tier memory (HOT/WARM/COLD), daily logs, semantic search (QMD), and lossless context management (Lossless...
⭐ 1· 359·2 current·2 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The skill's name/description match the deliverables in SKILL.md and the included script: it creates memory directories/files, checks/installs QMD and the Lossless Claw plugin via the OpenClaw CLI, and prints the openclaw.json changes to apply. Required binaries (bash, openclaw) are appropriate and proportional to the stated purpose.
Instruction Scope
Runtime instructions and the script are narrowly scoped to creating files under the provided workspace and prompting the user to modify openclaw.json. However, the AGENTS_TEMPLATE contains normative instructions such as "Don't ask permission. Just do it." and references heartbeat tasks (e.g., 'email checks, memory review') that imply autonomous access to external services. The skill itself does not provide credentials or automation for email/calendar access, but the template's phrasing encourages autonomous behavior and could lead to agents performing background checks if other parts of the system are configured to do so.
Install Mechanism
This is an instruction-only skill with a local bash script; there is no download-from-URL or archive extraction. Suggested installs (pip install qmd, openclaw plugins install @martian-engineering/lossless-claw) use package managers/CLI tools and are expected for these features. No high-risk install URLs are present.
Credentials
The skill requests no environment variables or secret credentials. It does suggest enabling heartbeat-driven checks that could require external service credentials (email/calendar) in normal agent operation — those credentials are not requested by the skill, which is appropriate, but users should be mindful to only provide such secrets to trusted plugins or agents.
Persistence & Privilege
always:false and no automatic modifications to global agent config are performed by the script (it prints config changes for the user to apply). The script writes only under the chosen workspace and is idempotent (it skips existing files). Autonomous agent invocation is allowed by platform defaults but not granted extra privilege by this skill.
Assessment
This skill appears to do what it says and is internally consistent, but review a few things before running: 1) backup any existing MEMORY.md/daily logs even though the script skips existing files (good defensive practice); 2) the AGENTS_TEMPLATE contains a blunt "Don't ask permission" instruction — if you want human-in-the-loop behavior, remove or edit that line so agents won't act without consent; 3) Lossless Claw is installed via the OpenClaw CLI (openclaw plugins install @martian-engineering/lossless-claw) — verify the plugin’s source if you don't already trust @martian-engineering; 4) heartbeat-driven tasks (email/calendar checks) will require separate credentials/configuration — the skill does not request them, so only grant such access through trusted, audited means; and 5) run the setup script in the intended workspace (not a shared or system directory) and verify file permissions. If you want extra assurance, inspect the plugin source and test in an isolated environment first.Like a lobster shell, security has layers — review code before you run it.
latestvk97da05zhbpsy2zhgyp9sjt3es83wvvn
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
Binsbash, openclaw