lessons-learned

Pass

Audited by Gen Agent Trust Hub on Feb 25, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The skill mines session history for lessons to persist, which creates a surface for indirect prompt injection if malicious instructions from the session context are saved to long-term memory.
  • Ingestion points: The entire conversation history of the current session.
  • Boundary markers: No specific delimiters are defined for content persisted to memory files.
  • Capability inventory: Write access to project-specific memory files, the global configuration (~/.claude/CLAUDE.md), and other skill files via a separate skill.
  • Sanitization: A human-in-the-loop approval step (Step 4) is mandatory before any content is persisted, effectively mitigating the risk of unintentional malicious injection.
  • [NO_CODE]: The skill consists entirely of markdown instructions and documentation; it does not contain any executable scripts, binaries, or automated package installations.
  • [SAFE]: The primary purpose of the skill is session persistence, which is managed through explicit user approval and clear quality guidelines.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 25, 2026, 04:26 PM