The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides examples of executing SQL queries against a PostgreSQL database using the postgres library to dynamically generate application paths.
Evidence: for await (const {id} of sqlSELECT id FROM products.cursor()) in SKILL.md.
[DYNAMIC_EXECUTION]: The documentation demonstrates the use of dynamic import() to fetch and execute JavaScript modules from remote URLs.
Evidence: import("https://my-app.example.com/chart.js") in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill describes an architecture that ingests untrusted data from an external database to define application routes, creating a potential surface for injection if the database content is compromised.
Ingestion points: Database query results (id field) used to construct route identifiers in SKILL.md.
Boundary markers: None present in the provided documentation examples.
Capability inventory: Database access (via postgres), dynamic module loading (via import()), and file system access for local imports in SKILL.md.
Sanitization: No explicit sanitization or validation of database-sourced strings is shown before interpolation into paths.
[EXTERNAL_DOWNLOADS]: The skill references and embeds image assets hosted on Observable's cloud infrastructure.
Evidence: src="https://observablehq.observablehq.cloud/oss-analytics/@observablehq/framework/downloads.svg" in SKILL.md.

observable-framework-embeds