The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references several external resources from the vendor's domain and well-known services:
Fetches the helix installation script from https://install.helix-db.com using curl | bash (REFERENCE.md).
Downloads and runs the ghcr.io/helixdb/enterprise-dev:latest container image for local instances (SKILL.md, REFERENCE.md).
Installs the @helix-db/helix-db npm package (version ^2.0.2) to evaluate TypeScript DSL queries (REFERENCE.md).
[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands:
The helix binary for project management and queries.
Container runtimes (docker or podman) for starting, stopping, and inspecting local database instances.
Node.js (v20+) is utilized to evaluate inline or file-based TypeScript DSL expressions.
[REMOTE_CODE_EXECUTION]: The helix query -e and --ts-file flags enable dynamic evaluation of TypeScript code via Node.js at runtime to generate database queries (REFERENCE.md).
[DATA_EXFILTRATION]: The skill accesses local sensitive configuration paths and environment variables to manage authentication and connectivity:
Reads credentials from ~/.helix/credentials and configuration from ~/.helix/config and ~/.helix/metrics.toml.
Accesses the HELIX_API_KEY environment variable and project-root .env files for cloud authentication headers.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes external data through query files:
Ingestion points: Raw JSON or TypeScript queries are loaded via the --file and --ts-file flags (SKILL.md, EXAMPLES.md).
Boundary markers: None identified; the CLI processes file contents directly.
Capability inventory: Subprocess execution of Node.js, container management via Docker/Podman, and network requests to local or cloud database endpoints.
Sanitization: The documentation notes that query validation is performed server-side by the Helix instance (SKILL.md).

helix-cli