uninstall-memo-hooks

Pass

Audited by Gen Agent Trust Hub on Jul 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill employs shell variable interpolation to construct Python scripts in modules/manifest.sh and modules/user-registry.sh. This pattern is susceptible to command injection if input variables such as mutation definitions or project paths contain malicious code sequences.
  • [DATA_EXFILTRATION]: The uninstaller modifies sensitive configuration files located outside the project directory, specifically ~/.claude/settings.json and ~/.claude/memo-flow/registry.json. These actions are used to deregister project hooks and update the environment's global state.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its reliance on the .claude/memo-flow/manifest.json file. The uninstall-memo-hooks.sh script deletes project files based on the mutations listed in this manifest without additional verification, creating a risk of arbitrary file removal if the manifest is tampered with.
  • Ingestion points: The project manifest file located at .claude/memo-flow/manifest.json (processed in uninstall-memo-hooks.sh).
  • Boundary markers: Not present; the script treats manifest entries as authoritative instructions for cleanup.
  • Capability inventory: Performs file deletion (rm -f) in uninstall-memo-hooks.sh and modifies global configuration via modules/settings-mutator.sh.
  • Sanitization: No verification is performed on the file paths or mutation IDs extracted from the manifest prior to the deletion or modification operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 7, 2026, 02:17 AM
Security Audit — agent-trust-hub — uninstall-memo-hooks