The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a Node.js script (render-elk.mjs) to convert JSON diagrams to SVG format. This execution uses a dynamic path (<diagrams-skill-dir>) and relies on the shell environment to run.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from external sources.
Ingestion points: Browsing and reading source code from external repositories (Workflow Step 1) and fetching external context from search results, blogs, and documentation (Workflow Step 2).
Boundary markers: None specified. There are no instructions for the agent to ignore or delimit embedded instructions within the source code or documentation it reads.
Capability inventory: The skill can execute shell commands, write files to the local disk, and clone remote repositories.
Sanitization: No sanitization or filtering logic is present to handle malicious content or hidden instructions within the analyzed codebase.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of a dependency via a third-party package manager (clawhub install diagrams) and involves cloning remote Git repositories for architectural research.
[DYNAMIC_EXECUTION]: The rendering process involves executing a script from an installed dependency. While the command is static, the path to the script is resolved at runtime based on the installation directory of the 'diagrams' skill.

architecture-research