The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides documentation on using @effect/platform for powerful system interactions like file system access (FileSystem) and command execution (Command). These features represent a potential vulnerability surface if an agent using this skill processes untrusted external data without proper validation.
Ingestion points: Data ingestion via fetch (documented in references/core-concepts.md and references/sched-scheduling.md) and FileSystem.readFileString (documented in references/plat-platform.md).
Capability inventory: Instructions for Command.make to execute shell processes and FileSystem.writeFile for file modification (documented in references/plat-platform.md).
Sanitization: The documentation consistently promotes the use of effect/Schema for strict data validation and provides detailed guidance on sanitizing inputs (found in references/schema-basics.md and references/schema-advanced.md).
[EXTERNAL_DOWNLOADS]: The documentation includes instructions for installing official @effect/* and @opentelemetry/* packages from public registries.
Evidence: Installation commands like npm install effect @effect/schema @effect/platform are provided throughout the getting started guides. These target well-known, established libraries in the TypeScript ecosystem.
[OBFUSCATION]: The skill uses Base64 and Hex encoding within code examples for educational purposes to demonstrate the library's ability to decode such formats.
Evidence: references/schema-advanced.md contains examples using Schema.StringFromBase64 with the input Zm9vYmFy (decodes to "foobar") and Schema.StringFromHex with a hex sequence. These are benign and intended for developer education.

effect-ts