Memory Hygiene

AdvisoryAudited by Static analysis on Apr 30, 2026.

Overview

No suspicious patterns detected.

Findings (0)

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.

What this means

If followed automatically or too casually, the user could lose all Clawdbot vector memories.

Why it was flagged

This is a destructive shell command that removes the entire LanceDB vector-memory directory. It is related to the skill purpose, but the artifact does not describe confirmation, backup, dry-run, or recovery steps before deletion.

Skill content
**Wipe:** Clear all vector memory
```bash
rm -rf ~/.clawdbot/memory/lancedb/
```
Then restart gateway: `clawdbot gateway restart`
Recommendation

Require explicit user confirmation before wiping, back up the memory directory first, and prefer an audit or selective-delete workflow when possible.

ConcernHigh Confidence
ASI10: Rogue Agents
What this means

A scheduled agent job could keep deleting and rebuilding memory each month, potentially removing useful memories without the user noticing.

Why it was flagged

The skill instructs creation of a persistent scheduled job that repeatedly performs a destructive wipe and reseed of memory. The automation is disclosed and purpose-aligned, but it lacks per-run approval, removal instructions, or rollback guidance.

Skill content
## Monthly Maintenance Cron

Set up a monthly wipe + reseed:

```
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"
}
```
Recommendation

Only add the cron job after an explicit user request, include instructions to disable it, and add backup/confirmation steps before each wipe.

What this means

Sensitive facts stored in memory may be recalled into future agent context, reused incorrectly, or exposed to later tasks.

Why it was flagged

The skill works directly with persistent vector memory and suggests storing account, credential-location, and contact facts. It also warns not to store OAuth URLs or tokens, which reduces but does not eliminate sensitivity.

Skill content
memory_recall query="*" limit=50 ... ✅ Store:
- Important facts (accounts, credentials locations, contacts)
...
❌ Never store:
- OAuth URLs or tokens
Recommendation

Store only minimal non-secret facts, avoid actual credentials or highly sensitive locations unless necessary, and review memory contents periodically.