Adityasagar
v1.0.0Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...
⭐ 1· 279·0 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.md, hook handlers, and scripts all implement a self-improvement / learnings capture workflow (logging to .learnings/, injecting reminders at bootstrap, detecting command errors, and scaffolding extracted skills). These capabilities match the described purpose. Note: there is a minor metadata/name inconsistency — the registry listing shows different names/slugs/owner ids (Adityasagar / aditya vs the package files which use 'self-improving-agent' / 'self-improvement') and the SKILL.md points to a GitHub repo; this is likely benign but worth verifying you are installing the intended package.
Instruction Scope
Runtime instructions stay within the stated purpose: create .learnings/ files, optionally copy/enable an OpenClaw hook, and run helper scripts. The hook handlers only inject a virtual reminder file into bootstrapFiles. The activator and error-detector scripts only print reminders (error-detector reads CLAUDE_TOOL_OUTPUT to detect failures). The extract-skill.sh script does write scaffolding to a local ./skills directory (and has checks to avoid absolute or ../ paths). Overall the instructions do not attempt to read unrelated secrets or exfiltrate data, but be aware that enabling hooks or adding scripts to user-level config will cause these scripts to execute with your user privileges.
Install Mechanism
No install spec in registry; installation instructions reference standard GitHub clone and an optional ClawdHub install. No downloads from obscure hosts or URL shorteners; scripts are local shell scripts. The code files are present in the package (hooks and scripts) but there is no remote code fetch at runtime beyond optional git clone from GitHub.
Credentials
The skill declares no required environment variables or credentials. The error-detector script reads CLAUDE_TOOL_OUTPUT (a runtime tool-output env var used by the agent tooling), which is expected in this context and not a secret. There are no requests for AWS/SSH/GitHub tokens or other unrelated credentials.
Persistence & Privilege
always:false and agent autonomous invocation is allowed (the platform default). The skill offers an optional hook that the user must copy/enable explicitly; enabling hooks will run the included scripts on events. The skill does not attempt to modify other skills or global agent configuration automatically (scripts perform writes only when run and extract-skill.sh explicitly protects against absolute/.. paths).
Assessment
This skill appears to implement exactly what it claims (reminders, simple hooks, logging templates, and a small helper to scaffold extracted skills). Before installing: (1) Confirm the package origin — the package files reference a GitHub repo and the registry metadata/slug names differ, so verify you're installing the intended repo. (2) Review the scripts locally (activator.sh, error-detector.sh, extract-skill.sh) before enabling hooks — hooks will execute those scripts with your user privileges. (3) If enabling PostToolUse hooks, prefer project-level configuration rather than global (user-level) to limit scope. (4) If you plan to run extract-skill.sh, prefer --dry-run first to see what it would create. If you want, I can list the exact lines to inspect in each script or produce a short checklist for safe installation.Like a lobster shell, security has layers — review code before you run it.
latestvk97a6x00gsdms1vr1y43tewwm1829tnb
License
MIT-0
Free to use, modify, and redistribute. No attribution required.