The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute the Bazel CLI for building, testing, and running compiled binaries (e.g., bazel build //..., bazel run //src:main). These commands are standard for a build system utility and operate within the expected scope of the skill's purpose.
[EXTERNAL_DOWNLOADS]: The skill demonstrates how to use MODULE.bazel to fetch external dependencies from the Bazel Central Registry, such as Abseil and GoogleTest. These references target well-known open-source projects.
[DATA_EXFILTRATION]: The skill includes configurations for remote execution and caching (--remote_executor, --remote_cache). While these features involve transmitting build metadata and artifacts to external servers, the provided examples use well-known services like Google Cloud Remote Build Execution (remotebuildexecution.googleapis.com) or generic placeholders.
[PROMPT_INJECTION]: The skill identifies an indirect prompt injection surface where the agent is instructed to process output from build queries using shell tools. • Ingestion points: Output from bazel aquery is piped to jq in the workflow. • Boundary markers: No explicit boundary markers or 'ignore' instructions are used for the tool output. • Capability inventory: The skill enables building and running arbitrary code defined in the project. • Sanitization: No sanitization of the queried build graph data is demonstrated. This is noted as an inherent risk of build system automation.

bazel