buck2-rule-basics
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- Command Execution: The tutorial involves running
buck2build commands and shell utilities likemkdirandcatto demonstrate build processes. These are standard operations for a developer tool tutorial and are intended to be executed in a dedicated workspace. - Dynamic Execution: The skill teaches the creation of Buck2 rules that generate and execute shell and Python scripts via the
ctx.actions.runAPI. This pattern is a core feature of the Buck2 build system for performing arbitrary build tasks. - Indirect Prompt Injection: The agent is designed to read local files such as
BUCKanduppercase.bzlto track user progress. This ingestion of local workspace data is a potential consideration for processing untrusted content, though it is limited to the tutorial's files and used to facilitate the interactive teaching flow. - Ingestion points: SKILL.md (reading
uppercase.bzl,BUCK,input.txtto assess state). - Boundary markers: Absent for local file reads.
- Capability inventory: File system access (
mkdir), file writing, and command execution viabuck2. - Sanitization: None specified for file content analysis.
- Vendor-Specific Configuration: The tutorial utilizes repository structures and naming conventions (e.g.,
fbcode//,ovr_config//) that are characteristic of the author's official development environment, representing standard usage for the tool.
Audit Metadata