Self-Improving Agent (Proactive + Self-Reflection)
ivangdavila/self-improving
Self-reflection + Self-criticism + Self-learning + Self-organizing memory. Agent evaluates its own work, catches mistakes, and improves permanently. Use when...
Runtime requirements
Runtime requirements🧠 ClawdisOSLinux · macOS · Windows
Security scan
Security ScanVirusTotalVirusTotalBenignView report →OpenClawOpenClawBenignhigh confidenceThe skill's declared purpose (local self-reflection and a persistent memory folder) matches the instructions: it creates and maintains files under ~/self-improving/ and updates workspace steering files; it does not request credentials or network installs and is internally consistent.Details ▾✓Purpose & CapabilityName/description (self-improvement, memory, heartbeat) align with the actions the skill instructs: creating a local ~/self-improving/ hierarchy, logging corrections, promoting patterns, and maintaining a heartbeat. Required binaries/env/configs are none, which is coherent for an instruction-only, local storage skill.ℹInstruction ScopeSKILL.md explicitly instructs the agent to read/write files in the user's home (~)/workspace (memory files, AGENTS.md, SOUL.md, HEARTBEAT.md). This is required for a persistent local memory system, but it grants the skill the ability to modify workspace steering files and create state in the user home; users should expect those file changes and review them. The instructions do not direct data to external endpoints or request secrets; boundaries.md forbids storing credentials.✓Install MechanismNo install spec and no code files — instruction-only. No downloads or binary installs are performed, so there is no install-time code execution risk.✓CredentialsThe skill requests no environment variables, no credentials, and no unusual config paths beyond ~/self-improving/ and optional workspace files. That matches its stated purpose of local memory management.ℹPersistence & Privilegealways:false (not force-included). The skill will create persistent files under the user's home and may autonomously invoke itself during normal agent operation (disable-model-invocation:false). Autonomous invocation + ability to read/write workspace files increases the blast radius compared to a purely ephemeral skill — appropriate for a memory system but worth user awareness.AssessmentThis skill is coherent: it implements a local self-improving memory by creating and maintaining ~/self-improving/ and adding small steering snippets to workspace files. Before enabling…