Security audit
Security checks for vulnerabilities and agentic risk
This is a straightforward HTTP client skill; it can send user-provided requests to external services, which is expected for its purpose but should be used carefully with secrets.
Install only if you need a general-purpose HTTP/API helper. Treat any URL, header, request body, bearer token, API key, or basic-auth value as data that may be sent to the chosen remote service, and prefer environment variables or a secret manager for real credentials.
The skill description does not warn users that entered data, headers, tokens, and request bodies may be transmitted to external HTTP endpoints. In a skill explicitly designed to contact arbitrary remote services, lack of disclosure increases the risk that users will provide sensitive information without realizing it will leave the local environment.
The activation phrases are extremely broad (e.g. 'HTTP', '发请求', '请求API') and can be triggered during normal conversation, causing unintended invocation of a network-capable skill. Because this skill can send arbitrary outbound requests with custom headers and bodies, accidental activation materially increases the chance of unintentional data transmission or misuse.
This example demonstrates sending an Authorization bearer token to an external endpoint and normalizes use of embedded credential material in skill examples. Even though the token is placeholder-formatted, the pattern encourages operators to paste real secrets into requests, and the client also stores request metadata in history, increasing exposure risk during use.
# 带认证的请求
auth_resp = client.request(HTTPRequest(
method="GET",
url="https://api.github.com/user",
auth="Bearer ghp_xxxxxxxxxxxx"
))
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# GraphQL
gql = client.graphql(
"https://api.github.com/graphql",
"{ viewer { login name } }"
)
print(f"GraphQL -> {gql.status}")
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# GraphQL
gql = client.graphql(
"https://api.github.com/graphql",
"{ viewer { login name } }"
)
print(f"GraphQL -> {gql.status}")
This example shows use of a bearer token for an external API and implicitly encourages users to provide live API secrets to the skill. Given that the skill supports arbitrary outbound requests and stores request history, this creates a substantial risk of credential disclosure, accidental exfiltration, or insecure operational use.
api = HTTPClient() api.request(HTTPRequest( method="GET", url="https://api.openai.com/v1/models", auth="Bearer sk-xxxxxxxxxxxx" ))