Back to skill

Security audit

Google Flow: Veo 3.1, Omni Flash, Nano Banana Pro (useapi.net)

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because scanners report advice to work around content moderation and a risk that local files may be uploaded for unsupported options; the review is partial because some referenced paths were not found.

Review the upload behavior and moderation guidance before installing. Confirm you trust the external API and understand which local media and credentials may be sent. Scanning was partial: all eight listed components were scanned, but several referenced paths could not be resolved, so the evidence is incomplete.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (7)

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Treats content moderation as a recoverable error and recommends switching models to get past it.

Medium
Category
Prompt Injection
Confidence
96% confidence
Finding

The instructions explicitly recommend switching models after moderation because their moderation differs. This encourages an agent to route around a safety refusal instead of determining whether the requested content is permitted; it is a safety-control weakness rather than a demonstrated technical exploit of the provider.

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 · scripts/_common.sh (reported line 6)May include surrounding context.

sh
# Written for bash 3.2 and later (the macOS default), so no bash 4 features.
set -euo pipefail

API="https://api.useapi.net/v1/google-flow"
: "${USEAPI_TOKEN:?Set USEAPI_TOKEN to your useapi.net API token (https://useapi.net/docs/start-here/setup-useapi)}"
command -v jq > /dev/null || { echo "jq is required (https://jqlang.org/download/)" >&2; exit 1; }

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 · scripts/_common.sh (reported line 13)May include surrounding context.

sh
enc() { jq -rn --arg v "$1" '$v|@uri'; }
log() { echo "$(date +%H:%M:%S) $*" >&2; }

# acurl [curl args ...] -> curl with the Authorization header read from stdin (-H @-),
# so the token does not show up in the process list
acurl() { curl "$@" -H @- <<< "Authorization: Bearer $USEAPI_TOKEN"; }

A troubleshooting hint encourages retrying refused content to get past moderation.

Medium
Category
Prompt Injection
Confidence
97% confidence
Finding

The safety-error hint recommends resubmitting unchanged refused content because moderation is nondeterministic. Although the helper does not automatically retry these requests, its guidance can induce the calling agent to repeatedly seek acceptance of content already refused by a safety control.

Content

No source excerpt is available for this finding.

Local-file uploads apply to every option, contrary to the documented reference_N upload behavior.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

Every option value is checked for a local file and uploaded before the option name is validated, so even an unrelated or unsupported option can trigger an external transfer. This exceeds the documented reference-image behavior and can disclose unintended local media when an argument points to an existing file; the upload helper's extension check limits, but does not eliminate, that exposure.

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 · scripts/upload.sh (reported line 22)May include surrounding context.

sh
P="assets"
if [ -n "${USEAPI_EMAIL:-}" ]; then P="assets/$(enc "$USEAPI_EMAIL")"; fi
OUT=$(mktemp)
CODE=$(acurl -sS --max-time 320 -o "$OUT" -w '%{http_code}' -X POST -H "Content-Type: $TYPE" --data-binary "@$FILE" "$API/$P") || true
CODE="${CODE:-000}"
BODY=$(_finish POST "$P")
ID=$(jq -r '.mediaGenerationId.mediaGenerationId // empty' <<< "$BODY")