Skillguard

ReviewAudited by ClawScan on May 1, 2026.

Overview

This skill appears purpose-aligned, but it uploads audited code to an external API and its optional installer can install a named skill after a safe verdict.

This looks coherent for a security-audit skill. Before using it, be comfortable sending audited source code to api.agentsouls.io, avoid auditing files that contain secrets or proprietary content unless you trust the service, and treat the safe-install wrapper as an installer that can modify your installed skills.

Findings (2)

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.

Medium

#ASI07: Insecure Inter-Agent Communication

What this means

Audited code is shared with an external service; this is disclosed, but local files could contain private source code or embedded secrets if the user selects them.

Why it was flagged

The script posts the selected skill or local file contents to https://api.agentsouls.io/api/audit for analysis.

Skill content

response=$(curl -sf -X POST "$API_URL" ... --arg code "$code" --arg source "$source" '{code: $code, source: $source}')

Recommendation

Only audit files you are comfortable sending to that provider, and use manual review or self-hosting for highly sensitive code.

Low

#ASI02: Tool Misuse and Exploitation

What this means

Installing a skill can change the agent’s future behavior, and the install decision depends on the external audit verdict unless the user chooses to inspect manually.

Why it was flagged

The helper installs the user-named skill automatically when the remote audit returns SAFE or LOW_RISK.

Skill content

SAFE|LOW_RISK) ... echo -e "✅ Skill looks safe. Installing..." ... exec clawhub install "$SKILL"

Recommendation

Review the audit output before relying on the wrapper, and avoid --force unless you already trust the skill being installed.