The Agent Skills Directory

[DATA_EXPOSURE]: The skill guides the agent to query official ImageKit documentation and community sources. It does not access sensitive local files or hardcode any credentials. All resources described align with the author's identity (imagekit-developer).
[PROMPT_INJECTION]: The skill uses assertive language in its description ('MANDATORY PRE-STEP', 'You MUST read') to ensure the agent follows the documentation search guidelines. This is standard instructional steering for specialized skills and does not attempt to bypass core safety guardrails or extract system prompts.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface where the agent processes external, untrusted content from the 'imagekit_community' source. While this content could theoretically contain malicious instructions, the skill does not possess sensitive capabilities (such as command execution, file system modification, or non-whitelisted network access) that could be exploited via this vector. The instructions focus on summarizing and citing information for the user.

search-docs