Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Openclaw Self Improve
v1.0.6Evidence-based and approval-gated self-improvement workflow for OpenClaw. Use when the user asks to make OpenClaw more powerful, optimize behavior, improve r...
⭐ 1· 100·0 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
Name/description match the implementation: scripts scaffold run artifacts, validate required files, and export JSON summaries for an improvement run. Required binaries (bash, git, date, grep, awk, python3) are reasonable for these tasks and no unrelated credentials or external services are requested.
Instruction Scope
Instructions operate on a target repo and create/validate files under <repo>/.openclaw-self-improve/<timestamp>/. They explicitly require the repo path (or fall back to /root/openclaw) and instruct the user to define validation gates (arbitrary commands). The skill does not itself execute the validation gate commands; it only records them. Risk note: because validation gates are free-form text, users must avoid supplying destructive commands or secrets in those fields.
Install Mechanism
No install spec; this is an instruction-only skill with bundled helper scripts. Nothing is downloaded or extracted from third-party URLs.
Credentials
No required environment variables, credentials, or config paths are declared. The scripts capture git commit/branch if the target is a git repo and otherwise work with 'n/a' defaults—this is consistent with the stated purpose.
Persistence & Privilege
The skill writes files into the repo you point it at. always is false and the skill is user-invocable. Caveats: --force can overwrite the run's standard artifact files (the script refuses to clobber a run dir that contains unexpected files), and the SKILL.md fallback to /root/openclaw means an agent that omits an explicit scope could operate on that path if it exists—confirm the target path before running.
Assessment
This skill appears coherent and safe for its declared purpose, but review these before installing/using: 1) Inspect the scripts yourself (init, validate, export); run init-improvement-run.sh with --dry-run first to confirm resolved values. 2) Explicitly pass the --repo path; do not rely on the default /root/openclaw unless you intend to operate on that repo. 3) Never put secrets or destructive shell commands into the validation-gate or other free-form fields. 4) Be cautious with --force: the script will refuse to overwrite a run dir containing unexpected files, but using --force can remove the standard artifact files. 5) Run validate-improvement-run.sh (and optionally export-improvement-run-json.py) locally to confirm outputs; verify run-info.json/summary.json for any inadvertently recorded sensitive info. If you need additional assurance, request a short security review of the scripts' behavior in your environment (e.g., run them in a non-production clone of the repo).Like a lobster shell, security has layers — review code before you run it.
latestvk97d26hwxb6akw26bazyq2k30s848bsa
License
MIT-0
Free to use, modify, and redistribute. No attribution required.