Skill Guardian
WarnAudited by ClawScan on May 10, 2026.
Overview
Skill Guardian matches its stated purpose, but its security scoring is largely placeholder code while it can automatically update other skills on a recurring schedule.
Review this skill carefully before installing or scheduling it. It does not show data theft or obvious malicious behavior, but its security claims are stronger than its implementation, and its auto-update workflow can change other installed skills. Prefer manual or dry-run mode until real vetting, exact update verification, and per-update approval are added.
Findings (4)
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.
You may believe a skill has been meaningfully security-vetted when the provided code did not actually evaluate the scanner result.
The shown vetting implementation ignores the captured vetter output and returns a fixed passing score, so the advertised trust score may not reflect real security checks.
# Parse vetter output (simplified)
return {
"trust_score": 80, # Placeholder - would parse actual output
"risk_flags": [],
"passed": True
}Require the vetter to return success, parse and display its real findings, fail closed on errors or nonzero results, and avoid assigning trust scores until real vetting is implemented.
A skill update can change future agent behavior without the user reviewing that specific update first.
The script can invoke the ClawHub CLI to update skills automatically based on registry trust scores, including immediate updates for high-trust entries.
if is_high_trust:
print(f"🌟 {skill_name}: High trust score ({trust_score}) - immediate update allowed")
...
subprocess.run(
["clawhub", "update", skill_name, "--version", info["latest_version"]],
check=True,
timeout=120
)Make dry-run the default, require explicit confirmation per update, verify the exact package and version before updating, and provide rollback instructions.
The skill can keep making changes to the skill registry/update state after the original setup, especially if added to cron.
The documentation recommends persistent scheduled execution, and the scheduled workflow includes applying updates.
🤖 **Auto-Scheduled** — Runs 1-2 times daily automatically ... Skill Guardian works best when run automatically 1-2 times daily. ... 3. 🔄 **Apply Updates** — High-trust (≥90) update immediately, others queued
Only enable scheduling if you want autonomous updates; otherwise run manually. If scheduled, log actions, review queued updates, and keep an easy way to disable the cron job.
A misidentified or insufficiently verified update could be queued and later applied to the user's skill environment.
The shown update detection relies on substring parsing of a global list output rather than exact package identity, source binding, or signed/versioned metadata before updates are applied.
result = subprocess.run(
["clawhub", "list"],
capture_output=True,
text=True,
timeout=30
)
...
if skill_name in line:
parts = line.split()
if len(parts) >= 2:
return parts[1] # Version columnUse exact package identifiers, owner/source checks, signed or API-provided metadata, changelog review, and explicit user approval before applying updates.