The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill's account connection workflow requires the user to provide raw login credentials for their X/Twitter account. The instructions explicitly list fields for username, email, password, and totp_secret to be collected by the agent.
[DATA_EXFILTRATION]: The collected raw credentials (including the password and TOTP secret) are transmitted via a POST request to an external third-party domain (https://twitr.sh/api/x-accounts/connect). Sending cleartext credentials to a third-party service instead of using delegated authentication (like OAuth) is a high-risk pattern for credential harvesting.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data such as post topic, tone, and text for drafting and publishing.
Ingestion points: The x_compose tool takes a topic and tone as input, while x_write takes the final text for publishing.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the interpolation points.
Capability inventory: The skill has the capability to perform network POST operations to publish content.
Sanitization: There is no evidence of sanitization or validation of the input content before it is processed or published.

x-twitter-publish