memory

SUSPICIOUS: the core recall/learn workflow is plausible, and its main infrastructure choices are generally consistent with a memory service, but the skill's actual footprint is broader than advertised. The undocumented git-based auto-install path weakens install trust, and the extensive Theory-of-Mind/persona features introduce disproportionate user profiling, background analysis, and optional third-party LLM data flow that do not cleanly belong in a simple 'query memory before scanning code' skill.