The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the pup CLI tool from the datadog-labs Homebrew repository. This is an official vendor resource provided by the skill author for the tool's intended functionality.
[REMOTE_CODE_EXECUTION]: The skill connects to an MCP server hosted on mcp.datadoghq.com, which is a well-known service domain for Datadog. This connection is used to fetch and search trace data.
[DYNAMIC_EXECUTION]: The skill generates and utilizes a local runner script (replay_runner.py) designed to import and execute specific entry points of the user's application code. This is the primary function of the skill and includes user confirmation steps for all code and environment changes.
[CREDENTIALS_UNSAFE]: The skill manages sensitive API keys (Datadog and LLM providers) but follows security best practices by recommending the use of .env files and environment variables instead of hardcoding or inlining credentials.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external traces to derive inputs for local execution. It mitigates the risk of processing malicious trace content by requiring the user to review and confirm the suggested input before replaying the code.

agent-observability-replay-trace