E.x.O. Installer
ReviewAudited by ClawScan on May 10, 2026.
Overview
This is a coherent installer and health-check skill, but it can run local commands, install global npm packages, enable OpenClaw tools, and optionally create scheduled monitoring.
Install only if you trust the E.x.O. npm/GitHub packages. Ask the agent to get confirmation before running install --all, update, internal clone, or cron setup, and review any optional Telegram or memory-tool configuration.
Findings (6)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Running install, update, or doctor commands may execute code on the local machine.
The CLI executes shell commands. This is aligned with an installer and health-check tool, but users should understand that running the skill can execute local npm/npx/package commands.
function runCmd(cmd, options = {}) { ... return execSync(cmd, {Only run these commands when you intended to install or manage E.x.O. tools, and review the package sources you are about to install.
The installed behavior depends on the current npm packages and their dependencies at install time.
The installer performs global npm installs based on package names from packages.json. This is central to the skill, but the artifacts do not pin exact versions.
runCmd(`npm install -g ${pkg.npm}`, { stdio: 'pipe' });Install only if you trust the E.x.O. package sources; consider checking package versions and npm/GitHub provenance before using install --all or update.
If used, internal cloning may access private repositories available to the user's GitHub account.
The skill documents an internal clone command that relies on GitHub organization access. This is disclosed and purpose-aligned, but it uses the user's GitHub privileges.
`exo internal clone` | Clone private repos (needs GitHub access)
Use internal clone only when you intend to access those repositories and understand which GitHub identity/credentials your environment will use.
A scheduled job may continue running health checks until the user removes or disables it.
The skill supports creating a scheduled health-check job. This is disclosed monitoring behavior, not hidden persistence, but it does keep running after setup.
`exo cron setup` ... creates an OpenClaw cron job that runs daily at 9am
Only run cron setup if you want ongoing monitoring, and confirm how to disable the job later.
Health-check details could be sent outside the local machine if Telegram alerts are enabled.
The docs disclose optional Telegram alerting. This may transmit health/status information to an external messaging service.
The `exo doctor` command aggregates these and can send Telegram alerts: ... `exo doctor --telegram`
Enable Telegram alerts only if you are comfortable sharing the reported health/status data with that Telegram bot/chat configuration.
If installed and configured, memory tooling may retain information for later agent use.
One installable package is an agent memory/RAG component. The installer itself does not show broad local indexing, but installing memory tooling can create persistent context stores.
"description": "Local RAG for AI agent memory — ChromaDB + sentence-transformers"
Review jasper-recall's own configuration for what it stores, where it stores it, and how to clear or disable retained memory.