The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill automates the creation and repair of verification recipes by ingesting repository-level data such as CLI exports, README files, and source code. This functionality presents a surface where malicious instructions embedded in a project's documentation or code could influence the behavior of the generated verification skill.
Ingestion points: Repository source code, CLI documentation, and observed runtime behavior as described in SKILL.md and references/project-skill.md.
Boundary markers: The instructions do not specify the use of delimiters or markers to isolate untrusted data from the generated instruction context.
Capability inventory: The skill has the capability to write new files to the filesystem (e.g., in .agents/skills/) and execute discovered repository commands.
Sanitization: There is no mention of sanitizing or validating ingested data before it is incorporated into generated scripts.
[COMMAND_EXECUTION]: The skill is intended to discover and execute project-specific commands to verify application state and readiness. This involves running external binaries and scripts found within the repository environment.

build-verifier