The Agent Skills Directory

Indirect Prompt Injection (HIGH): The skill's primary function is to fetch and display prompt content from an external API (Langfuse).
Ingestion points: The data.prompt field returned by the Langfuse API is printed directly to the console in scripts/langfuse-prompt-view.ts.
Boundary markers: No delimiters or instructions are provided to the agent to ignore any natural language instructions contained within the fetched prompt content.
Capability inventory: The skill uses npx tsx to execute TypeScript code via Bash, providing a vector for command execution if the agent is manipulated by injected instructions.
Sanitization: No sanitization or filtering is performed on the content retrieved from the external source.
Credential Exposure (HIGH): The script transmits sensitive credentials (LANGFUSE_SECRET_KEY) via Basic Authentication.
Evidence: In scripts/langfuse-prompt-view.ts, the Authorization header is constructed using environment variables and sent to LANGFUSE_HOST.
Risk: Since LANGFUSE_HOST (or LANGFUSE_BASE_URL) is configurable via environment variables, an attacker who can influence the environment can redirect these credentials to a malicious server.
Command Execution & Remote Dependency (MEDIUM): The skill relies on npx to execute the tool.
Evidence: SKILL.md defines the tool as npx tsx *scripts/langfuse-prompt-view.ts*.
Risk: This involves the runtime download and execution of the tsx package from the npm registry. While tsx is a common utility, using npx introduces a dependency on external registries and potential for version-shifting or package hijacking.

langfuse-prompt-view