The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's primary functionality is achieved through the execution of ae-cli commands to interact with the Agent Evaluation platform. The instructions provide specific command structures for managing datasets, evaluators, connections, and evaluation runs.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data for evaluation purposes, which represents a potential surface for indirect prompt injection.
Ingestion points: Local files are ingested via +import_dataset in references/datasets.md, and external traces are imported from AI Monitor via +import_ai_monitor_traces in references/integrations.md.
Boundary markers: The skill does not explicitly define prompt-level delimiters or boundary markers for the data being evaluated.
Capability inventory: The skill possesses the capability to read and write files and perform network operations through the ae-cli tool.
Sanitization: The documentation highlights that the ae-cli tool redacts secrets, credentials, and session tokens from command logs and error envelopes to prevent accidental exposure.
[SAFE]: The skill follows security best practices by requiring user confirmation for high-risk write operations (such as deletions or role changes) and by treating credentials as write-only secrets that are never logged or echoed.

ae-evaluation