Security audit
Security checks for vulnerabilities and agentic risk
The scanners rate this skill benign with high confidence: its findings mainly concern sending image prompts to an external service, while the other flagged patterns were judged unlikely to apply to its purpose.
Before installing, consider that image prompts are sent to veida.ai for processing. Avoid including confidential or sensitive details in prompts; the scanners do not report a privacy warning in the workflow.
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
mkdir -p .claude/skills/veida-image
curl -o .claude/skills/veida-image/SKILL.md \
https://raw.githubusercontent.com/hanshs474/veida-image-skill/main/SKILL.md
Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
mkdir -p .claude/skills/veida-image
curl -o .claude/skills/veida-image/SKILL.md \
https://raw.githubusercontent.com/hanshs474/veida-image-skill/main/SKILL.md
What it gives an agent: one image per generated client id, 1K and watermarked, roughly
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -X POST https://veida.ai/api/ai/generate \
-H "content-type: application/json" \
-H "x-anon-id: skill-$(openssl rand -hex 5)" \
-d '{
The workflow instructs an agent to send image prompts to an external provider without a privacy warning or guidance to exclude confidential information. Although external processing is inherent to this skill and the provider is named, a generic request for an image does not necessarily authorize disclosure of sensitive prompt details to that provider.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
`reason` is `anon_credits` (this client id is spent) or `anon_ip_daily` (this machine is done
for the day). Without checking for `wall` this surfaces as a confusing "no task id".
## Options