The Agent Skills Directory

[SAFE]: A comprehensive analysis of the skill's instructions, reference templates, and evaluation cases revealed no malicious code, unauthorized command execution, or network exfiltration patterns. All referenced files are local to the skill package and perform intended documentation and scaffolding functions.
[INDIRECT_PROMPT_INJECTION]: While the skill's primary function is to process user intent for generating AI prompts—a potential attack surface—it incorporates layered defenses. 1. Ingestion points: User requests are processed within SKILL.md to design system prompts and agent specs. 2. Boundary markers: The skill instructs the agent to define explicit output contracts and approval gates in all generated artifacts. 3. Capability inventory: The skill itself is purely instructional and lacks any internal capabilities to execute scripts, access the network, or write files. 4. Sanitization: Core instructions in SKILL.md explicitly forbid generating prompts designed for secret exfiltration, platform safeguard bypasses, or unauthorized access.

prompt-crafter