The Agent Skills Directory

[SAFE]: The skill serves as a best-practices guide for Go development and does not contain any malicious executable code or instructions. It outlines a standard technology stack including reputable services such as Stripe, Cloudflare, and Planetscale, and common open-source Go libraries like rueidis, bun, and testify.
[INDIRECT_PROMPT_INJECTION]: The skill defines a vulnerability surface by instructing the agent to process and modify files matching the **/*.go pattern. This allows the agent to ingest potentially untrusted project code which could contain embedded instructions. However, the guidelines themselves promote a high security posture through the following evidence chain:
Ingestion points: All files within the project matching the **/*.go pattern are targeted as context for the agent.
Boundary markers: No specific boundary markers are provided to separate project code from agent instructions, though this is typical for style guide skills.
Capability inventory: The agent's primary capabilities are reading, writing, and refactoring Go source code and tests.
Sanitization: The skill does not implement content sanitization, but it explicitly requires developers to write tests that validate security posture, such as assertOwnedBy and assertHiddenFromOtherTenants, which mitigates risks of accidental security regressions.

go