The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill includes external documentation links that have been flagged as malicious by reputation scanners.
Evidence: The URL https://jeffallan.github.io/claude-skills/skills/language/kotlin-specialist/ is blacklisted by automated security services.
[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by processing user-provided requirements to generate complex application logic and configurations without explicit sanitization or boundary markers.
Ingestion points: User-defined application requirements and architectural goals in SKILL.md.
Boundary markers: None present in the prompt instructions to delimit user input from agent logic.
Capability inventory: The skill workflow involves executing shell commands, writing files, and configuring server-side authentication (e.g., Ktor JWT setup).
Sanitization: No validation or escaping of user input is specified before interpolation into code templates.
[COMMAND_EXECUTION]: The workflow requires the agent to execute external CLI tools which could be abused if malicious instructions are injected into the build environment.
Evidence: SKILL.md explicitly instructs the agent to run detekt and ktlint during the validation phase.

kotlin-specialist