Structured episodic/semantic/procedural memory framework for agents.

What to check before installing or running this kit: 1) Verify the executable: the docs reference bin/memory-search but the manifest lacks that file. Inspect lib/search.sh and any bin wrapper expected by the docs. Do not run any 'memory-search' commands until you confirm which file is the CLI and that it is the intended code. 2) Inspect the scripts for network or remote-exec calls: open helpers/check-compaction.sh and lib/search.sh and grep for 'curl', 'wget', 'nc', 'ssh', 'scp', 'git', 'http', 'https', 'localhost', '127.0.0.1' or other external endpoints. The docs mention checking '/status' — find out what that refers to and whether the script will call an external service. 3) Avoid blindly running 'git pull' from an origin you don't control. If you clone from the public GitHub URL in README, verify the repository and its commit history first. Never run 'git pull' inside a skill directory from an unknown remote without review. 4) Backup your shell rc files before modifying PATH: the install docs recommend editing ~/.bashrc or ~/.zshrc. Save a copy and prefer adding an explicit alias or running the script via full path until you're confident. 5) Run locally in a sandbox first: test in an isolated environment (throwaway VM or container) to confirm behavior, especially compaction checks and any scripts that alter files or make network calls. 6) If you accept the kit, make the scripts executable yourself (chmod +x) and read them line-by-line. Look out for commands that write to locations outside the skill folder or that send data off-host. If you want, I can (a) scan the contents of helpers/check-compaction.sh and lib/search.sh for suspicious calls and summarize any network or destructive operations, or (b) produce a safe step-by-step sandbox install checklist you can follow.

Type: OpenClaw Skill Name: agent-memory-kit-2-1-0 Version: 1.0.0 The Agent Memory Kit is a structured framework designed to help AI agents manage long-term memory through episodic logs, semantic knowledge, and procedural guides. It includes a Bash-based search utility (`lib/search.sh`) that uses standard Unix tools like `grep`, `find`, and `jq` to query markdown files within the agent's workspace. The documentation and instructions (e.g., `SKILL.md`, `README.md`, `SEARCH.md`) are consistent with the stated purpose of improving context retention and surviving token compactions. No evidence of malicious intent, data exfiltration, or harmful prompt injection was found.