The Agent Skills Directory

Command Execution Logic: The skill provides instructions for utilizing command-line tools and APIs to manage cloud infrastructure, including the creation of database clusters and analytics graphs.
Evidence: SKILL.md includes specific management commands like create-db-cluster, modify-db-cluster, and create-graph.
Context: These operations are managed through explicit safety instructions requiring user confirmation and refusal lists for destructive actions like resource deletion.
Input Validation Mechanisms: A utility script is provided to validate user-supplied parameters before they are incorporated into commands or scripts.
Evidence: scripts/input_validator.py includes regular expression patterns for validating cluster identifiers, regions, and Amazon Resource Names (ARNs).
Context: This reduces the likelihood of executing malformed or unintended commands by ensuring parameters meet expected formats.
External Data Handling: The skill is designed to process data from external files and cloud storage, which is a potential consideration for indirect prompt injection.
Ingestion points: SKILL.md describes reading architecture requirements from JSON artifacts; references/migration.md and references/analytics-vs-database.md discuss importing graph data from S3.
Boundary markers: SKILL.md requires validating artifacts against a defined schema.
Capability inventory: The skill has the ability to execute CLI commands, write files to the local system, and run database queries.
Sanitization: Instructions in references/querying.md and scripts/graphrag_pipeline.py emphasize using parameterized queries (bindings) for all database interactions to prevent injection.
Secure Protocol Enforcement: The guidance explicitly requires the use of encrypted communication and authentication for all database access.
Evidence: references/security.md and references/connectivity.md emphasize TLS for transport, SigV4 signing for authentication, and encryption at rest.
Context: These instructions promote a least-privilege approach to network and data access.
Remote Tool Fetching: The skill provides instructions for downloading established graph analysis tools from well-known public repositories.
Evidence: references/connectivity.md provides a command to fetch the Apache Tinkerpop Gremlin console.
Context: The resource targets the official Apache software distribution server, which is an expected source for this technology.

amazon-neptune