Security audit
Security checks for vulnerabilities and agentic risk
This skill coherently helps generate music through MiniMax, with expected API-key use, external requests, and local audio-file output disclosed in the artifacts.
Install only if you are comfortable using MiniMax's API for generation. Prompts, lyrics, and your MiniMax API key are sent to the MiniMax service, and generated audio is written locally, so avoid submitting confidential or sensitive material unless you are authorized to share it.
Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.
)
try:
resp = requests.post(
api_url, json=payload, headers=headers, timeout=REQUEST_TIMEOUT
)
resp.raise_for_status()
Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.
if args.stream:
print("[info] Using streaming mode (direct response)...", file=sys.stderr)
try:
resp = requests.post(
API_URL, json=payload, headers=headers, timeout=REQUEST_TIMEOUT
)
resp.raise_for_status()
The skill instructs the agent to use environment variables, invoke a local script, write output files, and make API-backed generation requests, but it does not declare any explicit tool scope or permission boundaries. This creates a confused-deputy risk where an agent may exercise env, file-write, and network capabilities without transparent authorization, making accidental overreach or unsafe execution more likely.
The workflow mandates user-facing questions and example interactions entirely in Chinese, including required option-based prompts, without indicating that the agent should match the user's language or ask for preference. This creates a natural-language policy issue because the skill effectively enforces a specific language for interaction rather than offering a choice.
The required draft presentation, confirmation request, and post-generation feedback text are all specified in Chinese and presented as mandatory output formats. Because the skill does not provide an alternative based on user preference, it violates the language/locale policy criteria.
This file documents an external API endpoint, so outbound transmission is expected for the feature; however, it is still a real security-relevant behavior because user content and credentials are sent to an external service. In the context of a music-generation skill, the danger is not the existence of the endpoint itself but the absence of safeguards around what data is transmitted and how users are informed.
## Endpoint
`POST https://api.minimaxi.com/v1/music_generation`
## Auth
The documentation clearly instructs users to send prompts/lyrics to a third-party API and use an API key, but it does not warn that user-supplied lyrics or prompts may contain sensitive, proprietary, or personal data that will leave the local system. In a skill intended for interactive creative workflows, users may paste unpublished lyrics, private text, or confidential material, so the lack of a disclosure and consent boundary creates a real data exposure risk.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
from utils_audio import hex_to_bytes, save_bytes, download_url
API_URL = "https://api.minimaxi.com/v1/music_generation"
POLL_INTERVAL = 5 # seconds between status checks
MAX_WAIT_TIME = 900 # max 15 minutes for complex music
REQUEST_TIMEOUT = 180 # per-request HTTP timeout
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
)
try:
resp = requests.post(
api_url, json=payload, headers=headers, timeout=REQUEST_TIMEOUT
)
resp.raise_for_status()
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
if args.stream:
print("[info] Using streaming mode (direct response)...", file=sys.stderr)
try:
resp = requests.post(
API_URL, json=payload, headers=headers, timeout=REQUEST_TIMEOUT
)
resp.raise_for_status()
No suspicious patterns detected.