The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to use shell commands such as mkdir -p and mv to manage the lifecycle and placement of generated HTML artifacts within the user's project directory.\n
Evidence: SKILL.md contains explicit instructions for using these commands to ensure files are moved from agent sandboxes to the appropriate project workspace path.\n- [DYNAMIC_EXECUTION]: The skill includes a post-assembly validation script that executes an external binary to check the syntax of generated code.\n
Evidence: scripts/check-walkthrough.py uses subprocess.run(['node', '--check', ...]) to perform static syntax analysis on JavaScript code extracted from the assembled HTML file.\n- [INDIRECT_PROMPT_INJECTION]: The skill features a surface for indirect prompt injection by ingesting data from the user's workspace and specific context files.\n
Ingestion points: The agent is instructed to read CONTEXT.md (or a mapped context via CONTEXT-MAP.md) to extract term definitions for the walkthrough glossary in SKILL.md Phase 1 and Phase 4.\n
Boundary markers: No explicit delimiters or instructions to ignore embedded commands within the ingested context files are provided.\n
Capability inventory: The skill possesses file writing capabilities and command execution access via the shell and the check-walkthrough.py script.\n
Sanitization: No explicit sanitization or escaping of the ingested CONTEXT.md content is described before it is inserted into the HTML narration or glossary.

problem-description