The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses npx @leoflores/datadog-cli extensively across SKILL.md and reference files. This command downloads and executes code from the NPM registry at runtime. The package is not from a verified organization, and there is a discrepancy between the author name and the package maintainer, increasing the risk of supply chain compromise.
[COMMAND_EXECUTION]: In references/dashboards.md, the skill provides instructions for a 'Safe Dashboard Update Workflow' that involves constructing shell commands using variables like $WIDGETS, $DESCRIPTION, and $TEMPLATE_VARS. These variables are populated with content extracted from JSON files using jq. If the JSON content contains malicious shell characters or sequences, it could lead to command injection when these variables are interpolated into the npx command line.
[INDIRECT_PROMPT_INJECTION]: The skill functions by searching and tailing logs (logs search, logs tail) from Datadog, which are then processed by the agent. This represents a significant ingestion point for untrusted external data.
Ingestion points: npx @leoflores/datadog-cli logs search and logs tail (found in SKILL.md).
Boundary markers: No delimiters or instructions to ignore embedded commands are present in the prompt templates.
Capability inventory: The skill has the capability to write to the file system (via --output flag) and modify/delete cloud resources (via dashboards update and dashboards delete commands in SKILL.md).
Sanitization: There is no evidence of sanitization or filtering of the log content before it is returned to the agent context.

datadog-cli