Self-Improving Proactive Agent
Security checks across static analysis, malware telemetry, and agentic risk
Overview
This instruction-only skill is coherent and purpose-aligned, but users should understand that it creates persistent local memory and proactive follow-up state.
This skill appears safe to install if you want persistent local learning and proactive follow-up. Before using it, be comfortable with the agent maintaining local memory under `~/self-improving/` and `~/proactivity/`, and review any proposed changes to workspace steering files such as `AGENTS.md` or `HEARTBEAT.md`.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
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.
Stored corrections or preferences may shape how the agent behaves later, including across projects or future tasks.
The skill intentionally stores durable preferences and lessons that can influence future sessions. This is purpose-aligned, but persistent memory should be kept accurate and free of secrets.
Durable learning → `~/self-improving/` Use for: - confirmed preferences - reusable lessons - domain habits - project-specific patterns
Periodically review the local memory files, remove stale or incorrect entries, and avoid storing credentials, secrets, or highly sensitive information.
The agent may surface reminders, blockers, or prepared recommendations based on stored state rather than waiting for each prompt.
The skill includes proactive heartbeat behavior and recurring state checks. The artifacts bound this behavior to useful follow-up and do not show hidden background execution.
Good heartbeat checks - promised follow-ups that are due - stale blockers that may now be unblocked - deadlines or reviews approaching soon
Use the skill only if you want proactive follow-up behavior, and review or clear heartbeat state when a project ends.
If workspace steering is added, future agent runs in that workspace may inherit this behavior model.
Workspace steering files can influence future agent behavior across sessions. The artifacts frame this as integration guidance and also warn not to blindly duplicate rules.
Workspace integration guidance Add lightweight steering to: - `AGENTS.md` - `SOUL.md` - `HEARTBEAT.md` - `TOOLS.md`
Approve workspace integration explicitly, keep steering additions minimal, and review any changes to agent instruction files before accepting them.