The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute git -C <path> remote get-url origin where the path is derived from user input. This creates a potential command injection surface if the agent platform does not strictly sanitize shell arguments.
[COMMAND_EXECUTION]: The skill invokes a Node.js script ima_api.cjs from a sibling directory to perform API operations, which involves executing shell commands with dynamic JSON bodies.
[CREDENTIALS_UNSAFE]: The skill documentation notes that credentials (API keys or tokens) are automatically read from ~/.config/ima. Accessing sensitive credential files from the home directory is a high-privilege operation.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest external content such as URLs, files, and text snippets.
Ingestion points: User-provided links, uploaded files, and raw text entries in SKILL.md instructions.
Boundary markers: None explicitly defined in this skill; it relies on referenced core-rules.md in a sibling directory for safety protocols.
Capability inventory: File writing (via append_doc), network operations (via import_urls and ima_api.cjs), and shell command execution.
Sanitization: No explicit sanitization or escaping of the user-provided content is mentioned before it is processed or stored.

ybrain-capture