The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources (WeChat articles and comments), creating a surface for indirect prompt injection attacks.
Ingestion points: External WeChat article content and user comments are ingested via tools like wechat_mp_v2_article_detail and wechat_mp_v2_article_comments (referenced in SKILL.md and references/sandbase-api-map.md).
Boundary markers: The instructions do not specify any boundary markers or delimiters to separate the untrusted external content from the agent's instructions.
Capability inventory: The agent can execute tool calls using the sandbase_call_tool capability.
Sanitization: There is no mention of sanitizing, filtering, or escaping the external content before it is processed by the agent.
[NO_CODE]: The skill consists entirely of configuration and documentation files (Markdown) and does not include any executable scripts or binaries.

wechat-mp-research