openclaw-recovery-codex_skill

This skill helps Codex agents diagnose OpenClaw outages, discover environment details, and guide recovery steps across platforms.
  • Python

2.5k

GitHub Stars

2

Bundled Files

2 months ago

Catalog Refreshed

3 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill openclaw/skills --skill openclaw-recovery-codex

  • _meta.json309 B
  • SKILL.md6.6 KB

Overview

This skill performs discovery, diagnostics, and recovery guidance for OpenClaw Gateway and infrastructure used by Codex agents. It finds OpenClaw binaries, state and config locations, inspects status, ports, scheduled tasks, and common network helpers (like Tailscale). The skill outputs a structured recovery report and provides exact human commands for any action that must be run outside the agent.

How this skill works

The skill first discovers environment details: openclaw binary path, OPENCLAW_STATE_DIR, config path, OS and shell. It then runs non-destructive status checks (openclaw status, health, logs, port/listing tools, and service/task listings) and parses results for Gateway, channels, agents, memory/search, and security findings. When it detects issues it matches them to known patterns (Gateway unreachable, channel failure, port conflicts, config parse errors, webhook pipeline down, scheduled task mismatches, and permission problems). For safe fixes it can run read-only checks and for changes it emits exact ACTION_REQUIRED commands the operator should run.

When to use it

  • Gateway unreachable or ECONNREFUSED errors
  • Telegram/Discord/Signal channels reported OFF or erroring
  • Scheduled Tasks, services, or startup entries appear broken or stale
  • Webhook relay or Tailscale Funnel pipelines are not accepting traffic
  • openclaw status shows CRITICAL security or fts/memory degraded

Best practices

  • Always run full environment discovery; do not assume paths or config locations
  • Use read-only diagnostic commands first (openclaw status/health/logs) before making changes
  • Inventory all OpenClaw scheduled tasks and service units to avoid duplicate startup paths
  • Do not stop or kill the Gateway from an agent session; provide ACTION_REQUIRED commands for humans
  • When Windows JSON parse errors occur, run the BOM removal script rather than guessing edits

Example use cases

  • Detect and report an openclaw Gateway process that crashed and the exact restart command for systemd, launchctl, or schtasks
  • Identify a channel token misconfiguration using openclaw status --deep and recommend token rotation or network checks
  • Find port conflicts where multiple PIDs bind the Gateway port and list PIDs plus exact netstat/lsof commands
  • Discover missing SQLite fts5 causing degraded search and recommend fts installation steps
  • Locate stale scheduled tasks pointing to old install paths and provide commands to list them for human removal

FAQ

No. It runs safe, read-only diagnostics. For modifications it outputs exact ACTION_REQUIRED commands the human should run.

Do I need prior OpenClaw knowledge?

No. The skill discovers config and state dynamically and gives clear, actionable recovery steps for Windows, macOS, and Linux.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational
openclaw-recovery-codex skill by openclaw/skills | VeilStrat