The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from demo scripts and data model files, constituting a surface for indirect prompt injection.
Ingestion points: Reads files from demo/ and data/ directories as specified in Step 1.
Boundary markers: No explicit delimiters or instructions are used to separate user data from agent instructions.
Capability inventory: Generates configuration artifacts and provides templates for Elasticsearch API calls.
Sanitization: No sanitization or validation of the ingested file content is performed before processing.
[COMMAND_EXECUTION]: The skill provides structured templates for administrative API calls (PUT, POST, GET) intended for configuring Elasticsearch clusters.
[DATA_EXPOSURE]: Accesses project-relative paths and references a specific absolute file path /tmp/ddp-evals/diagnostic-test/dt-soct/dt-soct-current-state.json in its evaluation suite.

weave-train