Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks.
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· 118·0 current·0 all-time
by@lintqiu
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (log learnings, detect errors, promote learnings) align with the files and scripts provided. No unrelated credentials, binaries, or surprising requirements are requested. The included hooks and scripts directly support logging, reminders, and extracting learnings into skill scaffolds.
Instruction Scope
SKILL.md instructs copying a hook into OpenClaw hooks and creating a .learnings/ workspace that will be injected into sessions. That's expected for a "self-improvement" skill, but it does mean the skill injects reminder context into agent sessions (workspace/prompt injection). The scripts only emit reminder text or create local skill scaffolds; they do not exfiltrate data. Review that you want reminder injection into every session you enable and prefer project-level installs over global user-level if you want to limit scope.
Install Mechanism
There is no automated install spec; installation is manual (git clone or copying the hook). The package bundles scripts and hook handlers in the repo rather than downloading arbitrary binaries at runtime. This is a lower-risk install model, but installing or enabling hooks will cause agent-run code to execute (activator/error-detector scripts and the handler), so inspect scripts before enabling.
Credentials
The skill declares no required env vars or credentials. The error-detector script reads CLAUDE_TOOL_OUTPUT (an agent-provided env var) which is reasonable for detecting command failures but is not declared in requires.env — this is expected in OpenClaw/Claude contexts. No secrets or unrelated credentials are requested.
Persistence & Privilege
always:false (not force-included). However SKILL.md guides enabling hooks at user-level (~/.claude/settings.json or ~/.openclaw) which would cause the hook scripts and virtual-file injection to run for all sessions for that user. That is intentional for reminders but you should enable at project scope if you want to avoid global persistence.
Assessment
This skill appears to do what it says: inject lightweight reminders and provide helpers to log and promote "learnings". Before enabling it, review the included scripts and hook handler (they are short and readable). Prefer enabling hooks at the project level rather than user/global level if you want to limit injection to specific workspaces. If you run extract-skill.sh, be aware it will create files/directories under the current workspace (it includes checks to avoid absolute paths or '..' but still writes locally). No credentials are requested, but as with any hook that alters agent bootstrap/context, only enable it if you trust its content and the referenced repository.Like a lobster shell, security has layers — review code before you run it.
latestvk972zyg785jc9ac1arcfxxv8j5833269
License
MIT-0
Free to use, modify, and redistribute. No attribution required.