The Agent Skills Directory

[PROMPT_INJECTION]: The skill provides example code for a RAG pipeline and an automated evaluation system that are susceptible to indirect prompt injection due to lack of input sanitization.\n
Ingestion points: Untrusted data enters the context via the query parameter in the answer_question function and the test_set data in the evaluate_rag function.\n
Boundary markers: The provided prompt templates do not demonstrate the use of delimiters or specific instructions to help the model distinguish between instructions and external data.\n
Capability inventory: The examples showcase how to perform LLM calls and automated evaluation logic based on input data.\n
Sanitization: No evidence of sanitization, escaping, or validation of the external input is present in the provided snippets.\n- [EXTERNAL_DOWNLOADS]: The documentation provides standard instructions for installing official project dependencies from public registries.\n
Fetches the official @langtrase/typescript-sdk package from the public NPM registry.\n
Installs the langtrace-python-sdk package from the public PyPI registry.\n
Provides a Docker command to pull and run the official langtrace/langtrace-client image from Docker Hub.

langtrace