The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses python -c to check for specific library versions and utilizes package managers like pip, uv, or poetry to install necessary dependencies.
[EXTERNAL_DOWNLOADS]: The skill configures the environment to send telemetry data to the vendor's ingestion endpoints at https://ingest.maple.dev and https://ingest.eu.maple.dev.
[DATA_EXFILTRATION]: As a tracing tool, the skill is designed to transmit agent conversation transcripts, tool call arguments, and metadata to an external backend. It includes a content=False configuration option to allow users to disable the transmission of PII or prompt content while maintaining metadata visibility.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the agent's message history to generate tracing spans.
Ingestion points: The _messages function in maple_haystack.py processes ChatMessage objects containing user and assistant content.
Boundary markers: None are present; the tracer serializes all provided message content for export.
Capability inventory: The skill writes a local Python file (maple_haystack.py), installs packages, and performs network operations via the OTLPSpanExporter.
Sanitization: The skill uses json.dumps to serialize message components and tool results before export.

maple-agent-tracing-haystack