Self improving
v1.0.0Self-reflection + Self-criticism + Self-learning + Self-organizing memory. Agent evaluates its own work, catches mistakes, and improves permanently. Use when...
⭐ 0· 128·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The name/description (self-reflection, memory, learning) align with the instructions and manifest: it creates a local memory directory (~/self-improving/), logs corrections, and updates local workspace steering files (AGENTS.md, SOUL.md, HEARTBEAT.md). No unrelated credentials, binaries, or external services are required.
Instruction Scope
SKILL.md instructs the agent to read/write files under ~/self-improving/ and to add snippets to workspace files (AGENTS.md, SOUL.md, HEARTBEAT.md). That is coherent with the purpose but is an important scope decision: the skill will persistently read/write local files and edit workspace config files (non- destructive edits are emphasized). Users should expect these changes and review them.
Install Mechanism
This is instruction-only (no install spec, no code to download). The only install-like action is an optional, explicit user-consent step to run 'clawhub install proactivity' during setup; that would perform a network install only if the user agrees. There are no hidden downloads or extract steps in the skill itself.
Credentials
No environment variables, credentials, or external tokens are requested. The skill stores user preferences and corrections locally; boundaries.md explicitly forbids storing secrets/credentials. The requested access (file read/write in the user's home and workspace files) is proportional to the feature, but it means local persistent storage of user data.
Persistence & Privilege
The skill creates and uses persistent local state (~/self-improving/) and loads memory.md on every session. It is not force-enabled (always: false). Persistent local storage is expected for a memory skill, but users should be aware that learned items live on disk and will be read automatically by the agent.
Assessment
This skill is internally consistent and does what it says: it will create ~/self-improving/, log corrections/preferences, and update workspace steering files to make the agent behave more proactive. Before installing: (1) review and back up your workspace (AGENTS.md, SOUL.md, HEARTBEAT.md) because the skill suggests non‑destructive edits there; (2) understand that the agent will persistently read/write files in your home directory and load memory.md on sessions — avoid putting secrets or sensitive personal data into these files; (3) the only network action is the optional 'clawhub install proactivity' which will run only if you explicitly agree — decline it if you don't want additional network installs; (4) test in a disposable workspace or VM if you want to verify behavior first; and (5) if you ever ask the agent to 'forget everything', verify the exported backup before accepting deletion. Overall the footprint is proportionate to the stated purpose, but treat the created local memory as data you own and audit.Like a lobster shell, security has layers — review code before you run it.
latestvk973g5rf050emfdd5rqr8f22pn833w1m
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🧠 Clawdis
OSLinux · macOS · Windows