Back to skill

Security audit

Blog to Avatar Video

Security checks for vulnerabilities and agentic risk

Overview

This skill transparently helps create avatar videos through the revid.ai API, with ordinary third-party API and privacy considerations.

Install only if you are comfortable sending the blog URL, avatar image URL or character ID, and render metadata to revid.ai under your REVID_API_KEY. Avoid private/internal articles or personal images unless you have permission and accept that third-party processing.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (7)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding

The skill includes executable shell examples and polling logic but does not declare any explicit tool scope or allowed tools. In an agent environment, that omission weakens policy boundaries and can let the skill invoke shell/network-capable tooling without transparent restriction, increasing the chance of unintended command execution or data egress.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill processes user-supplied blog URLs and avatar image URLs by sending them to the external revid.ai service, but the description does not clearly warn users about that transfer. This creates a privacy and consent issue because users may provide sensitive or internal URLs or personal images without understanding they will be disclosed to a third party.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The example payload fixes voiceId to aria-en-us and language to en-US, and the failure guidance says the default voice is a gendered female English voice. This can violate language/locale policy because the skill defaults to a specific language and persona rather than asking the user or clearly limiting the skill to an English-only use case.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 105)May include surrounding context.

bash
# 1. Create character
curl -s https://www.revid.ai/api/public/v3/consistent-characters \
  -H "Content-Type: application/json" \
  -H "key: $REVID_API_KEY" \
  -d '{ "name": "Maya", "imageUrl": "https://cdn.example.com/maya.jpg" }'

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This manifest fixes the voice language to "en-US", which imposes a specific language/locale choice without indicating that users can opt in, override it, or that the skill is region-specific. That is a natural-language policy concern because it forces a locale preference rather than offering a choice.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
70% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · examples/run.sh (reported line 15)May include surrounding context.

sh
'.source.url=$u | .avatar.url=$a' \
  "$HERE/blog-to-avatar.json")"

PID=$(curl -fsS https://www.revid.ai/api/public/v3/render \
  -H "Content-Type: application/json" -H "key: $REVID_API_KEY" \
  -d "$PAYLOAD" | jq -r .pid)
echo "pid=$PID"

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

This shell script performs outbound HTTP requests to revid.ai and transmits user-supplied inputs as JSON, while also using an API key from the environment. Although the purpose is inferable from the comments, there is no explicit warning or disclosure that the provided URLs and render status data will be sent to a third-party service.

Content

No source excerpt is available for this finding.