The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill involves processing external API specifications (OpenAPI, AsyncAPI) to generate source code. This introduces a potential surface for indirect prompt injection if an attacker-controlled specification is used, though this is the core intended functionality of the tool.
Ingestion points: API specifications are ingested from local files or remote URLs defined in the generators.yml configuration (e.g., via the origin field).
Boundary markers: The instructions do not specify explicit delimiters or markers to isolate the agent from the natural language content within the API specifications.
Capability inventory: The skill enables code generation, local file system writes, Docker execution, and network operations for publishing to registries (npm, PyPI, etc.) and GitHub.
Sanitization: There are no specific instructions regarding the sanitization or validation of untrusted strings (such as endpoint descriptions or metadata) from the input specifications.
[EXTERNAL_DOWNLOADS]: The skill references the installation of the fern-api CLI tool from the npm registry and the addition of a Model Context Protocol (MCP) server from https://buildwithfern.com/learn/_mcp/server. Both resources originate from the vendor's official and recognized domains.

fern-sdks