The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill identifies and accesses a sensitive local file path for storing user authentication tokens.
Evidence: Mentions that credentials and browser-extracted cookies are saved to ~/.bilibili-cli/credential.json (SKILL.md).
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user-generated content from an external platform, which could contain instructions designed to hijack the agent's behavior.
Ingestion points: The skill fetches external data through commands like bili video --subtitle, bili video --comments, and bili search (SKILL.md).
Boundary markers: There are no explicit instructions or delimiters provided to help the agent distinguish between its system instructions and the untrusted content fetched from Bilibili.
Capability inventory: The skill has the ability to perform account actions such as bili like, bili coin, bili triple, bili dynamic-post, and bili unfollow, which could be abused if an injection is successful (SKILL.md).
Sanitization: No sanitization, filtering, or validation of the Bilibili subtitles or comments is described before the data is processed by the agent.

bilibili-cli