The Agent Skills Directory

[SAFE]: The skill uses the official volcengine-python-sdk to manage cloud resources, which is expected behavior for its stated purpose as a Volcano Engine utility.
[SAFE]: Destructive operations such as deleting instances, indices, or documents require an explicit --confirm flag and are governed by instructions requiring user confirmation.
[PROMPT_INJECTION]: The skill processes data from external Elasticsearch/OpenSearch endpoints, creating a potential surface for indirect instructions if those endpoints contain untrusted data.
Ingestion points: scripts/data.py via search, doc_get, and info commands.
Boundary markers: Output is returned as raw JSON without explicit delimiters to protect the agent from embedded instructions.
Capability inventory: Includes cluster management (lifecycle) and data operations (CRUD/search) via scripts/control.py and scripts/data.py.
Sanitization: Data retrieved from the search engine is passed directly to the agent without filtering.

byted-cloudsearch