The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run ./mvnw validate or mvn validate to ensure the project is in a valid state before processing.
[COMMAND_EXECUTION]: Uses shell commands like grep and find to analyze project structure, dependencies, and database access patterns.
[EXTERNAL_DOWNLOADS]: Fetches the PlantUML executable (plantuml.jar) from the official project repository on GitHub for diagram rendering.
[EXTERNAL_DOWNLOADS]: Includes external configuration for C4 model diagrams from the plantuml-stdlib GitHub organization.
[REMOTE_CODE_EXECUTION]: Executes the downloaded plantuml.jar using the java -jar command to validate syntax and generate image files.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted codebase data.
Ingestion points: Java source code (.java), Maven/Gradle configuration (pom.xml, build.gradle), and SQL schema files (.sql) accessed via codebase search tools.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are used when reading project files.
Capability inventory: Significant capabilities including shell command execution (mvn, grep) and binary execution (java -jar).
Sanitization: There is no evidence of sanitization or validation of content retrieved from project files before it influences agent actions.

033-architecture-diagrams