memory-hygiene
Pass
Audited by Gen Agent Trust Hub on Mar 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the recursive shell command
rm -rf ~/.clawdbot/memory/lancedb/to delete application memory directories as part of its 'wipe' functionality. - [COMMAND_EXECUTION]: The skill provides instructions to create a persistent scheduled task using the
crontool to automate destructive maintenance and reseeding actions. - [PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by instructing the agent to parse the
MEMORY.mdfile and store its contents into the vector database. - Ingestion points: Content is parsed from
MEMORY.mdduring the reseed or automated maintenance process. - Boundary markers: Absent; there are no instructions to use delimiters or ignore embedded instructions when reading from the markdown file.
- Capability inventory: Shell command execution (
rm -rf), process management (gateway restart), and system configuration modification (config.patch). - Sanitization: Absent; the skill does not specify validation or filtering for the data being stored.
Audit Metadata