The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is designed to export telemetry data, including LLM inputs, outputs, and metadata, to Confident AI endpoints (https://otel.confident-ai.com). This behavior is the central functionality of the skill and targets the vendor's own infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted data during telemetry export.
Ingestion points: The skill captures application-generated content, such as user prompts and LLM responses, into attributes like confident.span.input and confident.span.output as described in references/span-attributes.md.
Boundary markers: The skill does not provide specific instructions or delimiters to the agent to prevent the interpretation of instructions that might be embedded in the telemetry content.
Capability inventory: Telemetry data is transmitted via network operations using an OTLPSpanExporter to the vendor's backend.
Sanitization: While the skill uses json.dumps for object serialization, it does not provide methods for sanitizing or filtering LLM-generated text captured in spans.
[EXTERNAL_DOWNLOADS]: The templates/confident_otel_setup.py file and documentation recommend the installation of standard OpenTelemetry packages (opentelemetry-sdk, opentelemetry-exporter-otlp-proto-http) from the public PyPI registry.

confident-otel