The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources and the local environment which could be manipulated to influence the agent's decision-making or behavior.
Ingestion points: Data is read from the opik-release-policy.yaml file and the Opik API (specifically experiment metadata, dataset items, assertion results, and trace reasons).
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the data as non-executable text when processing assertion reasons or inputs.
Capability inventory: The skill has access to Bash for command execution and Write for file modifications.
Sanitization: The instructions do not specify any sanitization or escaping of the strings retrieved from the Opik API before they are used in the analysis report.
[EXTERNAL_DOWNLOADS]: The skill depends on the opik Python SDK to communicate with the Comet.ml/Opik services. This is a legitimate vendor resource used for the skill's intended purpose.

opik-verify