Self Improvement Ai
v1.0.0Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...
⭐ 0· 103·0 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 skill's name/description (capture learnings, errors, promote learnings) matches the provided files: documentation, templates, a hook to inject reminders, and helper scripts to detect errors and scaffold extracted skills. The presence of OpenClaw/OpenClaw-hook handlers, activator/error-detector scripts, and extract-skill.sh is appropriate for this purpose. Note: the package metadata lacks a homepage/source URL even though SKILL.md refers to GitHub repos; that is a provenance gap to verify but does not contradict the skill's functionality.
Instruction Scope
SKILL.md instructs creating .learnings/, copying/enabling an OpenClaw hook, and optionally running included scripts. The instructions stay within the stated domain (logging, reminding, extracting skills). The error detector reads the platform-provided CLAUDE_TOOL_OUTPUT variable to detect command failures — expected for a PostToolUse hook. Nothing in the instructions directs data to external endpoints or asks the agent to read unrelated secrets or system-wide configurations.
Install Mechanism
There is no automatic install/download step; it's instruction-only with local scripts and hooks included. No network downloads or archive extracts are performed by the skill itself. The extract-skill.sh writes files under a relative ./skills directory (it blocks absolute paths and '..'), which is reasonable for a scaffolding helper — review before running.
Credentials
The skill declares no required environment variables or credentials. Runtime scripts reference CLAUDE_TOOL_OUTPUT (a platform-provided variable) and standard shell behavior; this is proportionate to detecting tool errors. No unrelated secrets or external-service keys are requested.
Persistence & Privilege
always is false and the skill is user-invocable. Hook enablement is explicitly opt-in (you must copy/enable the hook). The hook injects a virtual reminder at bootstrap and scripts run with the same permissions as the agent when invoked — expected behavior for a hook-based reminder system. The skill does not modify other skills' configurations or demand permanent elevated presence.
Assessment
This skill appears coherent and low-risk for its stated goal, but before installing or enabling hooks: 1) Verify provenance (SKILL.md references GitHub but registry/homepage are blank) and trust the author. 2) Inspect the three shell scripts (activator.sh, error-detector.sh, extract-skill.sh) and hook handler files yourself; they are small and readable. 3) Note that enabling the hook causes those scripts/handlers to run with the agent's permissions and the error-detector reads CLAUDE_TOOL_OUTPUT (platform-provided tool output) — acceptable for local error detection but avoid enabling in multi-tenant or untrusted environments without review. 4) If you run extract-skill.sh, run in a safe working directory (it prevents absolute paths and '..' but will create files under the current path). 5) If unsure, enable only the activator (reminder) and skip PostToolUse/error detection until you've validated behavior.Like a lobster shell, security has layers — review code before you run it.
latestvk973b64j84596hvrv3rxs5r0cs8362e1
License
MIT-0
Free to use, modify, and redistribute. No attribution required.