memory-hygiene
2 findings (1 high, 1 medium), mainly resource abuse, unauthorized tool use. Most recent scan 2026-02-06.
This skill presents MODERATE security risks primarily related to destructive operations without adequate safeguards. The core functionality (memory maintenance) is legitimate, but the implementation lacks critical safety mechanisms. The most concerning issues are: (1) destructive rm -rf commands without warnings or backups, (2) automated cron jobs that delete data without user confirmation, and (3) configuration changes without validation. While there is no evidence of malicious intent (no data exfiltration, no credential theft, no prompt injection), the skill could cause significant data loss through accidental execution or automation failures. The missing junk.py file and lack of metadata are minor concerns. The skill would benefit from adding backup mechanisms, user confirmations, and failure handling before being considered production-ready.
Scanner's overall assessment (LLM analyzer)
Second scanner: NVIDIA SkillSpector
Independent static run, no language model, over the same files as the Cisco scan above. How the two scanners compare across the whole corpus: Two scanners, one corpus.
What this skill does
Audit, clean, and optimize Clawdbot's vector memory (LanceDB). Use when memory is bloated with junk, token usage is high from irrelevant auto-recalls, or setting up memory maintenance automation.
- Package
- 1 file, 1.9 KB
- SKILL.md
- 228 words, 6 sections
- Code
- None. Instructions only.
SKILL.md covers: Memory Hygiene · Quick Commands · Config: Disable Auto-Capture · What to Store (Intentionally) · Monthly Maintenance Cron · Storage Guidelines
Findings
HIGH Destructive Command Without Safety Guardrails Resource Abusellm
The skill instructs users to execute 'rm -rf ~/.clawdbot/memory/lancedb/' which permanently deletes all vector memory data. This command is presented without warnings, confirmation prompts, or backup recommendations. If executed accidentally or maliciously triggered, it causes irreversible data loss. The instruction lacks safety mechanisms like backup creation, user confirmation, or rollback capability.
SKILL.md
**Wipe:** Clear all vector memory
```bash
rm -rf ~/.clawdbot/memory/lancedb/
```MEDIUM Automated Destructive Cron Job Without Safeguards Unauthorized Tool Usellm
The skill provides instructions to set up a monthly cron job that automatically wipes all memory data without user interaction or confirmation. The cron job executes 'Wipe ~/.clawdbot/memory/lancedb/' on a schedule (monthly at 4 AM), which could cause unexpected data loss if users forget about the automation or if the reseed process fails. There are no failure handling mechanisms, backup procedures, or notification requirements before deletion.
SKILL.md
cron action=add job={
"name": "memory-maintenance",
"schedule": "0 4 1 * *",
"text": "Monthly memory maintenance: 1) Wipe ~/.clawdbot/memory/lancedb/ 2) Parse MEMORY.md 3) Store key facts to fresh LanceDB 4) Report completion"
}What is in the package
| File | Kind | Size |
|---|---|---|
SKILL.md | doc | 1.9 KB |
External hosts referenced
github.com
Hostnames that appear in the skill's files. Referencing a host is not a finding; compare against the findings above and the skill's stated purpose.
Package contents recorded 2026-10-22. The scan above ran against the version current at scan time; the two can differ if the skill has since been updated.
What we checked
- Static analyzer. Pattern rules over SKILL.md and bundled files: known-bad commands, exfiltration URLs, secrets, obfuscation.
- Behavioral analyzer. Checks what the skill asks the agent to do against what its description claims it does.
- Llm analyzer. A language model reads the skill and reports intent-level issues the rules cannot express.
- Trigger analyzer. Looks for instructions that fire on conditions the user did not ask for.
- Meta analyzer. A second model pass that validates, deduplicates and re-scores the other analyzers' findings.
What we did not check
- Runtime behaviour. The skill was not executed; findings come from reading its files.
- Anything fetched at install or run time, including remote scripts, package dependencies, and URLs the skill tells the agent to open.
- Content the analyzers cannot read: images, binaries, and some non-English text.
- Known scanner bypass classes. See the methodology and limitations page.
Sources
- Repository: https://clawhub.ai/skills/memory-hygiene
- Raw scan result: memory-hygiene-scan.json
- Badge for your README:
- Request a re-scan