Strict Self-Improving Agent (Rule of 3)
ReviewAudited by ClawScan on May 1, 2026.
Overview
The skill appears to do what it says—record and review lessons for future sessions—but its optional hooks and persistent memories can influence later agent behavior.
Install this if you want persistent self-improvement notes. Before enabling hooks, review the scripts and choose project-level scope when possible. Avoid logging secrets, credentials, or sensitive transcript details, and require explicit human approval before promoting learnings into core agent instruction files.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Private details or bad instructions written into learning files could be reused later by the agent.
Stored learnings can become future agent context or be promoted into workspace guidance, so inaccurate or sensitive entries could influence later sessions.
Context is injected from workspace files at session start ... In-session: Log to `.learnings/` ... Cross-session: Promote to workspace files
Review and redact learning entries before promotion, and keep human approval required for changes to SOUL.md, AGENTS.md, TOOLS.md, or other persistent guidance.
If enabled, the hooks run automatically in the agent environment, although the provided scripts only emit reminder text or inspect command output for error patterns.
The skill documents optional event hooks that run local shell scripts automatically after prompts or Bash tool use.
Hook scripts run with the same permissions as Claude Code ... Scripts only output text; they don't modify files or run commands ... All scripts are opt-in
Enable hooks only if you want automatic reminders, prefer project-level configuration, and review script contents before adding them to user-level hook settings.
Cross-session sharing may expose prior session context or send learnings to another agent if the user permits those tools.
The documentation describes cross-session transcript access, messaging, and sub-agent spawning as part of sharing learnings.
sessions_history Read transcript from another session ... sessions_send Send message to another session ... sessions_spawn Spawn a background sub-agent
Use cross-session tools only for sessions that should share context, and avoid sending secrets or sensitive transcript excerpts as learnings.