The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of child_process.spawnSync in scripts/statechart.mjs and scripts/test/statechart.test.mjs to execute other local Node.js scripts and the Java-based TLA+ TLC model checker.
[DYNAMIC_EXECUTION]: The script scripts/tlc-to-chart.mjs dynamically loads a JavaScript mapping file using the import() function. The path to this file is resolved from command-line arguments, which allows for the execution of arbitrary JavaScript code if the argument is manipulated.
[DYNAMIC_EXECUTION]: The test suite in scripts/test/statechart.test.mjs includes functionality to compile C source code at runtime using cc and subsequently uses the LD_PRELOAD environment variable to perform library injection. This is used to mock system calls such as readlink and rename to test the skill's handling of specific environment conditions.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external Mermaid statechart files (.mmd) and TLA+ specifications (.tla), which serves as an attack surface for indirect prompt injection.
Ingestion points: The skill reads and parses content from .mmd and .tla files.
Boundary markers: There are no explicit boundary markers or instructions to the agent to ignore potentially malicious content embedded in the labels or comments of the processed files.
Capability inventory: The skill can execute shell commands via subprocesses and perform file system writes.
Sanitization: While the skill uses specific parsers for its data formats, it does not sanitize natural language strings (like state labels or comments) that are subsequently presented to the agent.

statechart