Outclaw Setup

OutClaw setup: plugin inventory, connect outreach channels (Leadbay/LeadClaw, Gmail, Calendar, Slack, LinkedIn, WhatsApp, Calendly), capture the user's profi...

Plain-language checklist before you install or run this skill: - Understand the mismatch: the skill metadata declares no required env/config/install, but the runtime instructions will run shell commands, install third‑party plugins, and write to ~/.openclaw and memory files. That means the agent will perform real system actions and fetch code at runtime. - Review the source: inspect the actual scripts referenced (shared/scripts/kb_ingest.py, kb_page.py, memory_log.sh, inventory.py, etc.) and the listed plugin sources (e.g., arun-8687/linkedin-cli) so you know what will be installed and executed. Prefer installing only from auditable repos/releases. - Limit credentials: only connect accounts you control and are comfortable granting access. Prefer OAuth flows where you review scopes. Never paste sensitive callback URLs or long-lived secrets into untrusted chat channels. - Prefer manual review/run: if you want to be cautious, ask the agent to print the exact commands it would run instead of executing them, then run them yourself in a controlled environment (or sandbox/VM). - Backup state: snapshot ~/.openclaw and any memory files before running the wizard so you can recover if something is written incorrectly. - Ask the vendor for clarity: request that the skill declare required env vars, config paths, and an explicit install provenance for any plugins it will fetch. If they can't supply auditable install sources, treat the installs as higher risk. - Consider isolation: run initial setup in an isolated or disposable environment if you intend to test plugin installs. If you want, I can extract every command the SKILL.md would run and produce a checklist of exact scripts and external endpoints to inspect before you proceed.