Security audit
Security checks across malware telemetry and agentic risk
Keep is a broad but openly documented memory/context plugin that stores conversations and indexes workspace content; I found no artifact-backed deception or exfiltration.
Install only if you want an agent memory system that can persist conversation history and index workspace files. Review the OpenClaw indexPaths/indexExclude settings, choose local providers if you do not want content sent to model APIs or keepnotes.ai, use KEEP_NO_SETUP if you do not want automatic tool-config hooks, and keep dependencies updated.
capture_output=True,
timeout=60,
)
subprocess.run(
[npm, "run", "build"],
cwd=str(plugin_dir),
check=True,tmp = f.name
try:
subprocess.run([editor, tmp], check=True)
new_content = Path(tmp).read_text()
except (subprocess.CalledProcessError, KeyboardInterrupt):
typer.echo("Editor exited abnormally, no changes saved", err=True)tmp = f.name
try:
subprocess.run([editor, tmp], check=True)
new_content = Path(tmp).read_text()
except (subprocess.CalledProcessError, KeyboardInterrupt):
typer.echo("Editor exited abnormally, no changes saved", err=True)**The practice is `keep_prompt(name="reflect")`.** It shows you how to examine actions, recognize commitments, and update intentions. Run it. Follow the steps. Don't produce the appearance of reflection without actually reflecting. **Three tools:** `keep_flow` (all operations), `keep_help` (documentation), `keep_prompt` (practice prompts). ``` keep_prompt(name="reflect") # The practice
65/65 vendors flagged this skill as clean.
Detected: suspicious.dangerous_exec, suspicious.exposed_secret_literal, suspicious.prompt_injection_instructions