The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input via the $ARGUMENTS field, which is expected to contain Lodash code or function names.
Ingestion points: Input is ingested through the $ARGUMENTS variable in SKILL.md.
Boundary markers: There are no explicit boundary markers or instructions to ignore embedded commands within the processed code.
Capability inventory: The skill is allowed to use Bash, Read, Grep, and Glob tools.
Sanitization: No sanitization or escaping of the user-provided function names is performed before the agent is instructed to incorporate them into shell commands.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute Node.js scripts that inspect the local node_modules environment. These scripts resolve package paths and query exports to verify function availability.
[DYNAMIC_EXECUTION]: The workflow involves generating and running Node.js scripts via node --input-type=module -e to dynamically import and check entries from es-toolkit. This script generation uses templates where the agent is expected to insert values derived from user input.

migrate