Security audit
Security checks for vulnerabilities and agentic risk
This skill is a straightforward Gemini text-to-speech wrapper that sends user-provided text to Google's Gemini API and writes a WAV file locally.
Install only if you are comfortable sending the text, style prompts, and related request data to Google's Gemini API using your API key. Avoid using it with secrets, regulated data, or private documents unless your policy allows that, and monitor Google API quota or billing when selecting preview models.
Skill selects an external model or provider that may use a different account or billing plan than the operator expects. Undisclosed model switches can cause unexpected cost or quota consumption.
Fallback model selection:
scripts/gemini_tts.sh -m gemini-2.5-pro-preview-tts "Long-form narration example"
Skill selects an external model or provider that may use a different account or billing plan than the operator expects. Undisclosed model switches can cause unexpected cost or quota consumption.
Switch model examples:
scripts/gemini_tts.sh -m gemini-2.5-pro-preview-tts "Your text here"
scripts/gemini_tts.sh -m gemini-2.5-flash-preview-tts "Your text here"
Skill selects an external model or provider that may use a different account or billing plan than the operator expects. Undisclosed model switches can cause unexpected cost or quota consumption.
scripts/gemini_tts.sh -m gemini-2.5-pro-preview-tts "Your text here"
scripts/gemini_tts.sh -m gemini-2.5-flash-preview-tts "Your text here"
The README tells users to provide a Gemini/Google API key and use a hosted REST API, but it does not clearly warn that any text passed to the skill will be transmitted to Google's external service. In an agent setting, this omission can lead to accidental disclosure of sensitive prompts, PII, internal reports, or user content to a third party, especially if operators assume the skill is local-only.
The skill declares shell-capable behavior and lists required executables (curl, jq, base64, ffmpeg) but does not define any explicit tool scope such as permissions or allowed-tools. In an agent environment, this weakens least-privilege controls and can allow broader shell execution than the skill actually needs, increasing the blast radius if the script or its inputs are abused.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
ATTEMPT=1
HTTP_CODE=""
while (( ATTEMPT <= MAX_ATTEMPTS )); do
HTTP_CODE=$(curl -s -w "%{http_code}" -o "$TMPFILE" \
--connect-timeout 10 --max-time 60 \
-X POST "$API_URL" \
-H "Content-Type: application/json" \