The Agent Skills Directory

[Indirect Prompt Injection Surface]: The skill instructs the agent to read and evaluate repository files (such as third_party/v8/README.md) and upstream sources (such as denoland/rusty_v8 tags and V8 source files). Processing external data creates a potential indirect prompt injection surface if untrusted code or documentation content contains instructional text designed to alter agent behavior.
Ingestion points: Reading local files like third_party/v8/README.md and evaluating remote repositories/tags (denoland/rusty_v8).
Boundary markers: No specific delimiters or instructions to ignore nested directives are defined when processing these files.
Capability inventory: The skill calls local scripts via python3 and utilizes the GitHub CLI (gh).
Sanitization: The skill does not explicitly describe automated validation or content sanitization steps for the ingested data.
[Local Script and Command Execution]: The workflow specifies executing repository-internal scripts, such as .github/scripts/rusty_v8_bazel.py, along with running unit tests and checking GitHub CI workflows via gh. While standard for automated software release pipelines, executing local code requires that the repository environment and script logic are fully verified and restricted to the intended operations.

update-v8-version