The Agent Skills Directory

[SAFE]: The skill provides stylistic instructions and layout rules for various user interfaces (Telegram, CLI, Web). Its purpose is to maintain design consistency and visual quality within the project.
[COMMAND_EXECUTION]: The skill documentation references internal project CLI tools such as paw lab and paw api. These tools are used for simulating Telegram interactions and reviewing API outputs, which is consistent with the skill's role in refining user experience and does not indicate malicious intent.
[PROMPT_INJECTION]: The skill includes instructions to generalize user feedback into repository conventions. While this involves processing external input, the provided loop includes verification steps (treating feedback as a 'candidate rule' and requiring manual generalization), which mitigates risks associated with indirect prompt injection.

pawrrtal-taste