Auto-Updater (Gateway)

AdvisoryAudited by Static analysis on Apr 30, 2026.

Overview

No suspicious patterns detected.

Findings (0)

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.

What this means

A bad or breaking upstream skill update could be installed automatically across the user's agent environment.

Why it was flagged

The workflow updates every installed skill. The artifact does not show allowlisting, version pinning, changelog review, or rollback before accepting upstream changes.

Skill content
`clawdhub update --all`
Recommendation

Prefer report-only or approval-gated updates, review changelogs, allowlist what may update, and keep a rollback or backup plan.

What this means

The assistant runtime could change or restart on its own schedule, causing disruption or adopting unexpected behavior after an update.

Why it was flagged

The scheduled job message can direct the agent to mutate the core Clawdbot installation and restart the gateway. That is high-impact automation without a per-run confirmation step in the recommended command.

Skill content
`if Clawdbot has an update available, apply it and restart; then run clawdbot doctor --non-interactive`
Recommendation

Use report-only mode unless you explicitly want unattended self-updates; require manual approval for Clawdbot updates and restarts.

What this means

Skill names, versions, paths, errors, and operational details from update reports could be sent to the wrong Telegram recipient.

Why it was flagged

The recommended delivery command contains a concrete Telegram recipient ID rather than a placeholder. If copied, update reports may go to an unintended account.

Skill content
`--deliver --channel telegram --to "2095290688" ... report what changed`
Recommendation

Replace the recipient with your own verified destination or remove delivery; avoid sending verbose logs to external channels.

NoteHigh Confidence
ASI10: Rogue Agents
What this means

The updater may keep running daily until the cron job is disabled or removed.

Why it was flagged

The skill intentionally creates a recurring Gateway cron job and can wake it immediately. This is purpose-aligned, but it persists after setup.

Skill content
`clawdbot cron add ... --cron "0 4 * * *" ... --session isolated ... --wake now`
Recommendation

After setup, verify the cron job list and know how to pause or delete the job before enabling unattended updates.

What this means

Updates will run with whatever permissions the logged-in ClawHub account has.

Why it was flagged

The workflow relies on a logged-in ClawHub CLI session to update skills. This is expected for the stated purpose, and the artifact does not show credential logging or unrelated use.

Skill content
`clawdhub login --workdir /home/xabo/clawd --dir skills` and `clawdhub whoami`
Recommendation

Confirm the logged-in identity, use least-privilege credentials where possible, and avoid sharing logs that may reveal account or environment details.