Security audit
Security checks for vulnerabilities and agentic risk
This skill is a coherent local TTS debugging guide, with no hidden execution or malicious behavior found.
Before installing, confirm the configured TTS base URL is a server you trust. Treat uploaded voice recordings as sensitive data, get clear consent, and make sure the server has access controls and a deletion or retention policy.
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
Examples:
- Chinese output: write `一 二 三`, not `123`
- English output: write `one two three`, not `123`
This rule matters because the TTS model can go weird or read digits badly when fed raw numerals.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
Check the basics against the actual configured host:
curl http://127.0.0.1:8000/health
curl http://127.0.0.1:8000/v1/models
The documentation explicitly states that uploaded voice recordings and associated metadata are persisted on disk, but provides no warning about retention, access control, consent verification, or deletion handling. Because voice recordings are sensitive biometric and personal data, normalizing this storage pattern without privacy safeguards can lead operators to deploy a system that retains user data longer than intended or in an insecure manner.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
示例：
curl http://127.0.0.1:8000/v1/audio/speech \
-H 'Content-Type: application/json' \
-X POST \
-d '{
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
强制非流式：
curl http://127.0.0.1:8000/v1/audio/speech \
-X POST \
-H 'Content-Type: application/json' \
-d '{
The example response returns absolute filesystem paths to stored audio and metadata files, which unnecessarily discloses server-internal path structure to clients. Such path disclosure can aid reconnaissance by revealing deployment layout, usernames, mount points, or storage conventions that make subsequent attacks easier.