OpenClaw Recovery (Codex)
v1.0.0OpenClaw Gateway recovery and infrastructure diagnostics for Codex agents. Use when Gateway is unreachable, Telegram/Discord/Signal channels are down, Schedu...
⭐ 0· 308·1 current·1 all-time
byShunsuke Hayashi@shunsukehayashi
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name and description match the content: all files and instructions focus on discovering OpenClaw binaries, state/config, ports, services, and common failures. There are no unrelated credentials or unrelated third-party services requested in metadata.
Instruction Scope
SKILL.md and detect-env.sh instruct the agent to run many local system diagnostic commands (openclaw, lsof/ss/netstat, systemctl/launchctl/schtasks, tailscale, find under $HOME). This is consistent with diagnostics, but the skill will read local config files and search the user's home directory for openclaw.json — ensure the agent is permitted to inspect those files. The SKILL.md explicitly marks destructive actions as REPORT ONLY and provides safe vs report-only guidance, which reduces risk.
Install Mechanism
Instruction-only skill with no install spec. The included shell script is small and non-obfuscated. No downloads, archive extraction, or external install URLs are used.
Credentials
No environment variables or credentials are required by the registry metadata. The instructions reference OpenClaw-specific env vars (OPENCLAW_STATE_DIR, OPENCLAW_CONFIG_PATH) and common system tools; these are proportionate to a diagnostics/recovery utility and do not request unrelated secrets.
Persistence & Privilege
The skill does not request always:true, does not modify other skills' configurations, and includes no persistent installers. It does include a Node one-liner to remove a BOM from a config file (a write operation) but that action is documented and intended as a human-invoked recovery step.
Assessment
This skill appears to do what it says: local discovery and recovery for OpenClaw. Before installing, consider: (1) the agent will run or suggest many system commands and will search your home directory for openclaw config files — only allow it if you consent to that inspection; (2) destructive actions (killing services, changing permissions, creating/deleting tasks) are listed as REPORT ONLY — verify the agent actually restricts itself and ask a human to run risky commands; (3) the skill assumes system tools and the openclaw binary exist (the metadata didn't list required binaries) — it may fail if tools are missing; (4) review the included detect-env.sh and the Node BOM one-liner yourself before allowing execution. If you don't fully trust the skill's source, run the scripts manually in a controlled environment instead of granting autonomous execution.Like a lobster shell, security has layers — review code before you run it.
latestvk975qt7v8z7tm85w7rmz175jys81zk51
License
MIT-0
Free to use, modify, and redistribute. No attribution required.