The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, specifically project Gradle files, version catalogs, and Kotlin source code as part of its primary workflow.
Ingestion points: Reads project structure, Gradle files (build.gradle.kts), version catalogs (libs.versions.toml), and Kotlin source code to determine the appropriate DI configuration (detailed in SKILL.md Step 0).
Boundary markers: The skill instructs the agent to strictly follow non-negotiable rules defined in references/rules.md to prevent unintended behavior.
Capability inventory: The agent has the capability to generate Kotlin code, modify build configurations, and suggest command-line operations.
Sanitization: While no explicit input filtering is described, the skill enforces strict structural templates and relies on the Koin Compiler Plugin for build-time verification.
[EXTERNAL_DOWNLOADS]: The skill references external repositories and services for dependency management and runtime analysis.
Evidence: Fetches version metadata from Maven Central (repo1.maven.org) and references the Kotzilla MCP endpoint (mcp.kotzilla.io).
Note: These resources are official vendor services or well-known public package registries, which are considered safe sources.
[COMMAND_EXECUTION]: The skill contains instructions for the agent to guide the user through command-line operations for setting up development tools.
Evidence: The skill mentions commands for adding MCP servers (claude mcp add) and configuring the Kotzilla SDK (detailed in references/kotzilla.md).
Note: The skill includes explicit security guidelines forbidding the agent from auto-executing these commands, ensuring that all environment changes require manual user approval and action.

koin