The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands using local runtimes such as python3 and node. It explicitly utilizes debugging tools like pdb, debugpy, and the --inspect flag for Node.js to observe program state. This is part of its core functionality to verify code behavior rather than relying on static analysis.
[DYNAMIC_EXECUTION]: The skill facilitates dynamic execution by instructing the agent to insert 'instruments' (e.g., print statements, console.log, or breakpoints) into existing source code. It then runs the modified code to capture real-time state. To mitigate potential side effects, the skill includes a mandatory rule (RULE-06) to remove all instrumentation and verify the code's original state via git diff before concluding.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it is designed to ingest and process untrusted data from the user's project, including source code, logs, and documentation.
Ingestion points: The agent reads local project files (SKILL.md body references) and documentation to generate explanations.
Boundary markers: No explicit delimiters or 'ignore embedded instructions' warnings are specified when the agent reads external file content.
Capability inventory: The skill has access to shell execution (python3, node), debugger tools, and platform commands (/run, /verify).
Sanitization: There are no explicit sanitization or filtering steps mentioned for the content read from the project files.

teach-me