The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection via path traversal in its file handling examples.
Ingestion points: Untrusted data enters the application through the params.name parameter in src/routes/api/files/[name]/+server.ts.
Boundary markers: Absent; the code lacks any delimiters or constraints to prevent the traversal of the directory structure.
Capability inventory: The code snippets in src/routes/api/files/[name]/+server.ts utilize Bun.file for reading and Bun.write for writing to the file system.
Sanitization: Absent; there is no validation or escaping performed on the user-provided file name before it is interpolated into a file path.

bun-sveltekit