The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references and fetches content from official vendor sites and well-known service registries to maintain documentation and examples.
Evidence: Fetches directory listings from the GitHub API at api.github.com/repos/alchemy-run/alchemy-effect to sync examples.
Evidence: Initializes git submodules from the official alchemy-run/alchemy-effect repository.
Evidence: Recommends installing standard NPM packages like alchemy and effect from public registries.
[COMMAND_EXECUTION]: The skill uses local CLI commands to manage the infrastructure development lifecycle, which is standard behavior for deployment tools.
Evidence: Executes bun alchemy deploy, bun alchemy dev, and bun alchemy destroy to manage cloud resources.
Evidence: Uses bun tsc -b and bun build:clean for project compilation and maintenance.
[INDIRECT_PROMPT_INJECTION]: As an infrastructure management tool, the skill naturally ingests untrusted configuration data which represents a baseline surface for indirect prompt injection.
Ingestion points: The agent reads alchemy.run.ts configuration files, package manifests, and environment variables.
Boundary markers: The skill does not explicitly define delimiters for user-provided configuration strings in its instructions.
Capability inventory: The skill utilizes file system access, network operations via the Effect HttpClient platform, and shell command execution through the alchemy CLI.
Sanitization: No specific sanitization or escaping logic for user-provided resource identifiers is detailed in the skill's instructions.

alchemy