The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to execute various shell commands and API requests to verify the state of the environment.
Evidence: Instructs the agent to perform GET and POST requests to Elasticsearch endpoints (e.g., /{index}/_count, /_inference/sparse_embedding/{id}) and references the execution of local scripts such as python3 scripts/inject-anomaly.py and python3 scripts/warm-elser.py.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes data from multiple external files that could contain malicious instructions.
Ingestion points: The skill reads from demo/{slug}-discovery.json, demo/{slug}-platform-audit.json, demo/{slug}-demo-script.md, data/{slug}-data-model.json, data/{slug}-ml-config.json, and demo/{slug}-current-state.json.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the content of these files as untrusted or to ignore any embedded directives.
Capability inventory: The agent possesses capabilities to execute shell commands, perform network operations via Kibana APIs, and write files to the deploy/ directory.
Sanitization: The instructions do not include steps to sanitize or validate the content of the files before they are processed by the agent.

finish-check