The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted data (interview transcripts) which provides a surface for indirect prompt injection attacks.
Ingestion points: The skill takes raw interview text and pre-existing codes provided by the user in SKILL.md.
Boundary markers: The skill recommends a mapping format ("原文" → [编码标签]), but lacks strict delimiters or instructions to ignore embedded commands within the processed transcripts.
Capability inventory: The skill uses the Write tool to create and modify Markdown files (coding_*.md, themes_*.md) in the project root and the user's home directory (~/Documents/research-memos/themes/).
Sanitization: No explicit logic is provided to sanitize or escape the input text before it is interpolated into the prompt or written to files.
[COMMAND_EXECUTION]: The skill automates file system operations using the Write tool to save analysis results.
The skill defines specific file naming conventions and paths, including the project root and ~/Documents/research-memos/themes/. While these actions are central to the skill's utility, the combination of processing untrusted text and writing to the filesystem is a documented security surface.

thematic-analysis