Memory System V2
v1.0.0Fast semantic memory system with JSON indexing, auto-consolidation, and <20ms search. Capture learnings, decisions, insights, events. Use when you need persistent memory across sessions or want to recall prior work/decisions.
⭐ 12· 4.6k·28 current·31 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
high confidencePurpose & Capability
The name/description (file-based semantic memory) lines up with the included bash CLI and jq usage. Required binary (jq) and brew install are appropriate. Minor inconsistencies: README/SKILL.md claim automatic features (auto-capture, automatic weekly consolidation) and integration hooks, but the shipped code provides only a manual CLI; some example paths/command names differ (examples show memory/memory-cli.sh or memory-cli.sh and refer to memory_search which is not a provided command).
Instruction Scope
Runtime instructions and docs instruct agents/users to read and search local memory files (MEMORY.md, memory/*.md), which is expected for a memory skill. However the SKILL.md and design docs promise 'auto-capture' and 'automatic' weekly consolidation; the implementation only provides a manual 'consolidate' CLI command and no background scheduler or hook. The integration snippet references a command ('memory_search') that does not exist in the codebase — this mismatch could cause agents to run incorrect commands or attempt to read unexpected paths. The CLI writes/reads files under $HOME/clawd/memory which will persist potentially sensitive data locally.
Install Mechanism
Install spec only pulls in jq via Homebrew (a well-known package source). No remote arbitrary downloads or extract steps present. The skill is mostly instruction + a single bash script, so installation risk is low.
Credentials
No environment variables or external credentials requested. The script uses $HOME to store files — expected for a local file-based memory system. No network endpoints or tokens are requested.
Persistence & Privilege
always:false (good). The skill persists data to $HOME/clawd/memory (index, daily logs, consolidated files). This is expected but means the skill will store potentially sensitive, persistent information on disk; there is no automatic cleanup or encryption. The skill does not modify other skills or system-wide settings.
What to consider before installing
This skill is a small, local bash-based memory system that only requires jq and writes to $HOME/clawd/memory — nothing in the code phones home or asks for credentials. However: (1) the docs claim automatic capture/consolidation and reference a 'memory_search' hook that is not implemented — automatic behavior is NOT present in the shipped script; (2) the CLI must be run manually (or you must set up your own cron/agent hooks) if you want scheduled consolidation/auto-capture; (3) it will persist anything you capture to disk in plain text, so do not store secrets or sensitive data there without additional protections; (4) consider running it in a limited account or sandbox first and review/patch the small documentation/command-name mismatches before integrating into agents. If you need true automatic capture/consolidation or stronger guarantees about uniqueness/encryption, request or implement those features before deployment.Like a lobster shell, security has layers — review code before you run it.
latestvk975kx9fnbn7bb1s44x63d7c8180ah2p
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🧠 Clawdis
Binsjq
Install
Install jq via Homebrew
Bins: jq
brew install jq