The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download and execute a CLI installation script directly from the vendor's GitHub repository.
Evidence: curl -fsSL https://raw.githubusercontent.com/semantius/semantius-cli/main/install.sh | bash in SKILL.md and references/cli-usage.md.
[REMOTE_CODE_EXECUTION]: Documentation includes script templates (Bun, Python, and Shell) intended for bulk data imports that utilize network operations and system calls.
Evidence: Webhook import scripts in references/webhook-import.md use fetch(), requests, and curl to send data to the platform API.
[COMMAND_EXECUTION]: The skill's primary function is to orchestrate shell commands using the semantius CLI to manage database schemas and records.
Evidence: Numerous examples of semantius call crud ... and semantius call cube ... throughout the reference files and evaluation sets.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to process and import untrusted external data (CSV/Excel files).
Ingestion points: Data ingestion occurs through CSV, Excel, and TXT files as described in references/webhook-import.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands within the processed data are provided.
Capability inventory: The skill utilizes shell command execution via the semantius CLI, file system writes (Bun.write, Bun.file), and network operations (fetch, requests, curl).
Sanitization: The provided script templates use standard JSON serialization (JSON.stringify, json.dumps) to ensure data is correctly encoded before being sent to the API.

use-semantius