The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes a workflow step to verify data flow by consuming messages from Kafka topics, which introduces a surface for indirect prompt injection if those messages contain malicious content.
Ingestion points: Data is brought into the agent's context through the mcp__confluent__consume-messages tool as described in SKILL.md (Phase 4: Verification).
Boundary markers: The skill instructions do not specify the use of delimiters or specific "ignore" instructions to wrap the external Kafka message data.
Capability inventory: The agent possesses high-impact capabilities, including the ability to create connectors, execute Flink SQL statements, and enable Tableflow materialization via the Confluent MCP server.
Sanitization: There are no explicit validation or sanitization steps defined for the content consumed from Kafka before it is processed by the agent.
[DYNAMIC_EXECUTION]: The skill dynamically generates and executes SQL statements and infrastructure configurations based on environment discovery and user-supplied parameters.
Evidence: SKILL.md (Phases 2 and 3) provides templates for the agent to assemble JSON connector configurations and Flink SQL statements (such as CREATE TABLE and INSERT INTO), which are then executed through mcp__confluent__create-connector and mcp__confluent__create-flink-statement.
[EXTERNAL_DOWNLOADS]: The skill references an external repository for the Confluent MCP server to facilitate tool setup and validation.
Evidence: SKILL.md contains a reference to the official Confluent repository at https://github.com/confluentinc/mcp-confluent.

confluent-cloud-cdc-tableflow