The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to interact with the shell to verify git status, determine the current branch, and identify the repository's root directory name. It also instructs the agent to check the system PATH for the presence of specific CLI tools like Playwright and Cypress to determine environment capabilities.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from an external file (skills/jt-find-skills/SKILL.md) and populates placeholders in a generated markdown file based on that data. This creates a potential surface for indirect prompt injection.
Ingestion points: Reads the content of skills/jt-find-skills/SKILL.md under the header "ค้น".
Boundary markers: The instructions lack explicit boundary markers or warnings to the agent to disregard any instructions that might be embedded within the ingested data.
Capability inventory: The skill has capabilities to read and write files, execute git commands, and check the system PATH.
Sanitization: No validation or sanitization is performed on the extracted skill names before they are written to the card-loop/board.md file, which may be processed by downstream skills like jt-card-gate.

jt-new-board