The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates fetching and processing content from external websites, which introduces a surface for indirect prompt injection if the retrieved data contains instructions that the agent might inadvertently follow.
Ingestion points: The skill utilizes impit.fetch, AsyncClient.get, and impit.get to retrieve response bodies from arbitrary URLs as described in references/nodejs.md and references/python-rust.md.
Boundary markers: No specific delimiters or instructions to ignore embedded content are provided for the agent when processing response data.
Capability inventory: The environment supports network operations (via impit) and shell command execution for package management and building.
Sanitization: The instructions do not mention sanitizing or validating external content before it is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to download packages from public registries and source code from external repositories.
Evidence: References standard installations using bun add impit and pip install impit. It also specifies fetching source code and patched libraries from https://github.com/apify/impit.git, https://github.com/apify/rustls.git, and https://github.com/apify/h2.git for Rust build processes.
[COMMAND_EXECUTION]: The skill includes instructions for running commands to manage dependencies and build the library.
Evidence: Workflow and reference files include commands for bun, pip, and cargo to install the Impit client and its related dependencies.

impit