The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and process data from external financial announcements and government portals which could contain malicious instructions.
Ingestion points: Workflow instructions in SKILL.md to pull latest announcements and the source map in references/idc-monitoring-indicators.md.
Boundary markers: The skill does not provide specific delimiters or instructions to treat external data as untrusted or separate from system instructions.
Capability inventory: There are no shell commands, file system writes, or network-capable scripts included in the analyzed skill files.
Sanitization: No sanitization, validation, or filtering logic is specified for the external data processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to access specific external URLs for gathering financial and regulatory data.
Resources: Official A-share information disclosure portal (cninfo.com.cn), operator procurement portals (b2b.10086.cn), and government websites (miit.gov.cn, nea.gov.cn).
Context: These links are official resources relevant to the skill's stated purpose of monitoring corporate and regulatory risk signals.
[NO_CODE]: The skill contains markdown documentation and template files; no executable scripts, binaries, or package management configurations were found.

risk-warning-catalysts