Security audit
Security checks for vulnerabilities and agentic risk
This is a coherent content-outline skill, with a privacy caveat that URL inputs are fetched through an external SkillBoss scraping service.
Install only if you are comfortable with URL references being processed by SkillBoss API Hub. Prefer pasting non-sensitive excerpts or using public URLs, and avoid private links, signed URLs, tokens in query strings, internal documents, or authenticated pages. Review the generated outlines for overly aggressive metaphors before posting.
The example phrasing explicitly includes violent language such as 'serial killer' as a style model for output. Because the skill directs the model to emulate these examples and use absolute, imperative language, it materially increases the chance of generating harmful or extremist-adjacent content inappropriate for general social-post assistance.
The skill instructs the agent to read an API credential from the environment and use it to fetch arbitrary user-provided URLs through an external scraping service. For a content-outline generation skill, this expands capability into networked data retrieval and third-party transmission, which creates unnecessary data-exfiltration and misuse risk if sensitive or internal URLs are supplied.
The skill tells the agent to send user-provided URLs to SkillBoss API Hub for scraping but does not require a user-facing disclosure that the URL and retrieved content will be transmitted to a third party. This can expose confidential links, tokens embedded in URLs, or sensitive page contents without informed consent.
The finding identifies the same outbound request construction, showing the skill packages data for external transmission to api.heybossai.com. In a content-idea skill, this is dangerous because it broadens the trust boundary and can send sensitive reference sources off-platform without meaningful safeguards.
SKILLBOSS_API_KEY = os.environ["SKILLBOSS_API_KEY"]
def fetch_url(url: str) -> str:
r = requests.post(
"https://api.heybossai.com/v1/pilot",
headers={"Authorization": f"Bearer {SKILLBOSS_API_KEY}", "Content-Type": "application/json"},
json={"type": "scraper", "inputs": {"url": url}},
The finding identifies the same outbound request construction, showing the skill packages data for external transmission to api.heybossai.com. In a content-idea skill, this is dangerous because it broadens the trust boundary and can send sensitive reference sources off-platform without meaningful safeguards.
SKILLBOSS_API_KEY = os.environ["SKILLBOSS_API_KEY"]
def fetch_url(url: str) -> str:
r = requests.post(
"https://api.heybossai.com/v1/pilot",
headers={"Authorization": f"Bearer {SKILLBOSS_API_KEY}", "Content-Type": "application/json"},
json={"type": "scraper", "inputs": {"url": url}},
The hardcoded external endpoint confirms dependency on a third-party service for scraping user-specified content. The risk is not the hostname alone, but that the skill normalizes sending arbitrary target URLs and resulting page content to an external processor without addressing privacy, authorization, or internal-resource access concerns.
def fetch_url(url: str) -> str:
r = requests.post(
"https://api.heybossai.com/v1/pilot",
headers={"Authorization": f"Bearer {SKILLBOSS_API_KEY}", "Content-Type": "application/json"},
json={"type": "scraper", "inputs": {"url": url}},
timeout=60,
No suspicious patterns detected.