The Agent Skills Directory

[COMMAND_EXECUTION]: The /fppt:viewer command executes local shell operations, including npm --prefix ppt-viewer run build and node server.mjs. This allows the agent to trigger build processes and maintain a local web server on port 9030. While these are functional requirements for the skill's purpose, they represent a vector for executing local code within the agent's environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted text data from user-supplied files to determine PPT structure, content, and layout.
Ingestion points: Untrusted data enters the agent context via files located in the work/input/001_项目名/ directory.
Boundary markers: The instructions do not define specific delimiters or "ignore embedded instructions" warnings for the processed input data.
Capability inventory: The skill possesses significant capabilities, including writing structured JSON files to the file system, generating SVG assets in work/assets/, and executing shell commands via the viewer tool.
Sanitization: There is no evidence of input validation, escaping, or filtering to prevent malicious instructions embedded in the input materials from influencing the agent's generation or execution logic.

proposal-generator