Supercharged Memory

Transforms the AI into a lifelong companion by capturing, storing, consolidating, and searching context across sessions automatically and securely.

This package appears internally consistent with a local memory system, but review these before installing: - Expect file writes: the agent will create and update files in your workspace (memory/, MEMORY.md, heartbeat/health files) automatically and silently per the protocols. If you prefer manual control, do not enable automatic session-start activation or run the setup steps by hand. - Avoid storing secrets: do not tell the agent passwords, credit cards, or tokens as 'memories'. The README/SECURITY docs explicitly advise this; follow that guidance. - Optional network features are explicit: the vector DB (Mem0 + Qdrant) and dashboard sync/Supabase are optional and documented. Only enable them if you understand the consequences (embedding API keys may be sent to an embedding provider; Supabase sync would replicate data to a remote DB). - Inspect scripts before executing: SETUP-PROMPT and the scripts will run filesystem operations. They appear well-written, but run them manually (line-by-line) or in an isolated test workspace if you want to audit behavior first. - Backup & encrypt: if memories will contain sensitive info, keep your workspace encrypted (FileVault/LUKS/BitLocker) and back up the memory directory safely. - Prompt-injection caution: the skill includes explicit defenses, but no defense is perfect. Be cautious when asking the agent to ingest arbitrary external documents; treat the memory files as sensitive and audit them occasionally. If you want a lower-risk trial, copy the skill into an empty test workspace and exercise it there (verify file creation, QMD fallback to grep, and optional upgrade flows) before adding it to your daily workspace.

Type: OpenClaw Skill Name: normieclaw-supercharged-memory Version: 1.0.3 The Supercharged Memory skill is a well-architected system for managing long-term agent context through local file storage and the QMD search tool. The bundle demonstrates strong security awareness, including explicit prompt injection defenses in SKILL.md, the use of restrictive file permissions (chmod 600/700) in SETUP-PROMPT.md, and a comprehensive SECURITY.md audit. The provided shell scripts (qmd-reindex.sh and memory-health-check.sh) are transparent, perform only local file and process management, and contain no indicators of data exfiltration, unauthorized network activity, or malicious intent.