The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes an external script (../../check-update.sh) during its initialization preamble. This script is located outside the skill's own directory, making its contents unverifiable and creating a dependency on the host environment's file structure.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes untrusted user-controlled files. Ingestion points: The skill reads docs/02-方案设计/PRD产品需求文档.md and docs/01-需求调研/MVP方案.md in Step 2 to guide its output generation. Boundary markers: There are no explicit instructions or delimiters to prevent the agent from following instructions that might be embedded within these documents. Capability inventory: The skill possesses Bash and Write capabilities, which could be exploited if an attacker successfully injects commands into the processed PRD or MVP files. Sanitization: No evidence of content validation or sanitization is present in the skill's processing logic.

pm-data