The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data which could potentially be used to influence agent behavior through indirect prompt injection.
Ingestion points: Data is collected from user responses via AskUserQuestion (Step 2.2) and read from a local file docs/02-方案设计/数据指标体系.md (Step 2.1).
Boundary markers: The skill interpolates external data directly into sub-agent prompts (e.g., {acquisition 相关数据}) and the final report without using delimiters or instructions to ignore embedded commands.
Capability inventory: The skill possesses the Agent tool (to spawn sub-tasks), Write (to create documentation), and Bash (used for basic file presence checks).
Sanitization: There is no evidence of data validation, escaping, or sanitization before the external content is processed or written to files.

pm-aarrr