The Agent Skills Directory

[Prompt Injection] (SAFE): No instructions to override agent behavior or bypass safety filters were found. The content is strictly technical Haskell guidance.\n- [Data Exposure & Exfiltration] (SAFE): No evidence of hardcoded credentials, sensitive file path access, or unauthorized network communication was identified. The reference to local documentation is benign.\n- [Obfuscation] (SAFE): The file contains plain text markdown with no encoded sequences, zero-width characters, or deceptive homoglyphs.\n- [Remote Code Execution] (SAFE): No remote scripts are downloaded or executed. References to build tools like Cabal or Stack are instructional and do not trigger subprocess execution.\n- [Indirect Prompt Injection] (LOW): The skill processes user queries about functional programming, which is an inherent input surface, but it lacks the necessary system-level write or network capabilities to escalate an injection into a significant threat.

haskell-pro