The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze distributed trace and span data from OpenSearch indices. This data, often originating from external applications, represents an attack surface where adversarial instructions could be embedded to influence the agent's reasoning during trace analysis.
Ingestion points: Accesses trace data in otel-v1-apm-span-* and service map data in otel-v2-apm-service-map-* via the opensearch-mcp-server.
Boundary markers: The instructions do not define specific delimiters or "ignore" warnings to distinguish between the agent's instructions and the content of the analyzed trace fields.
Capability inventory: The skill utilizes MCP servers for database interaction and web search, and provides shell command templates (curl) for manual execution. This combination allows for significant interaction with the environment.
Sanitization: There is no mention of filtering, escaping, or validating the content of trace fields (such as span attributes or exception messages) before they are processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill configuration downloads and executes MCP servers (opensearch-mcp-server-py, duckduckgo-mcp-server) using the uvx tool. It also instructs the agent to fetch PPL command documentation directly from the opensearch-project/sql repository on GitHub. These operations target the vendor's own infrastructure or established community utilities.

trace-analytics