The Agent Skills Directory

[SAFE]: The skill uses vendor-owned tool interfaces (sandbase_describe_tool, sandbase_call_tool) for all external data interactions. No unauthorized network calls or hardcoded credentials were identified.
[SAFE]: All operations are research-focused and read-only. The instructions explicitly prohibit taking actions on external platforms, and no persistence or privilege escalation mechanisms are present.
[INDIRECT_PROMPT_INJECTION]: The skill ingests content from external sources (Twitter, Reddit, Weibo, Google News), which presents a theoretical ingestion surface for indirect prompt injection.
Ingestion points: Responses from sandbase_call_tool containing social media posts, news articles, and trends.
Boundary markers: Instructions require the agent to cite sources, preserve attribution, and separate factual observations from interpretations, providing logical boundaries for external data.
Capability inventory: Restricted to tool-based data retrieval and internal synthesis; no file-writing, shell access, or administrative capabilities are exposed to the agent.
Sanitization: Not explicitly implemented, but the risk is mitigated by the limited scope of available tools and the requirement for evidence-backed synthesis.

crisis-monitor