Self Improving Agent
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· 92·0 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 (capture learnings, errors, promote to workspace) matches the included files and instructions: logging templates, promotion guidance, hook handlers, and helper scripts. Required env/config/credentials are none, which is appropriate for this purpose.
Instruction Scope
Runtime instructions are limited to creating/using .learnings files, injecting a virtual reminder at bootstrap, running small hook scripts that emit reminders, and scaffolding extracted skills. They do not instruct the agent to read unrelated system files, access credentials, or send data to external endpoints. The error-detector script reads the CLAUDE_TOOL_OUTPUT environment variable (expected for PostToolUse hooks) to detect error text — this is consistent with the declared purpose.
Install Mechanism
There is no automatic installer; installation is manual (git clone or clawdhub). No archive downloads, no remote arbitrary code fetch at runtime. The only remote reference is a GitHub repo (author-provided). Helper scripts live in the package and are intended to be run locally.
Credentials
The skill does not require credentials or sensitive environment variables. The scripts use CLAUDE_TOOL_OUTPUT (a runtime hook-provided variable) and standard local paths within the workspace; this fits the skill's function. No disproportionate secrets or unrelated env vars are requested.
Persistence & Privilege
always:false and normal autonomous invocation settings. Enabling the OpenClaw/Claude hooks allows the hook handler and scripts to run during agent lifecycle events with the agent's local permissions (expected for hooks). This is normal but means the enabled hook scripts can write to the workspace and create skills (extract-skill.sh writes to a relative ./skills directory).
Assessment
This skill appears to do exactly what it says: log learnings, inject lightweight reminders, and help scaffold extracted skills. Before enabling hooks or running scripts, verify you trust the source (the SKILL.md references a GitHub repo), inspect the scripts (activator.sh, error-detector.sh, extract-skill.sh) yourself, and run them in a controlled workspace if you want to be cautious. Note: enabling hooks gives those scripts the same local permissions as your agent (they can create/modify files under your workspace). If you only want reminders, use the activator and skip the PostToolUse hook; if you use extract-skill.sh, prefer --dry-run first. If you have any sensitive workspace files, avoid installing or enable hooks only in isolated projects.Like a lobster shell, security has layers — review code before you run it.
latestvk974926kzm5qs91tqxkghxw1ps83fk5g
License
MIT-0
Free to use, modify, and redistribute. No attribution required.