The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the AI agent to execute a local binary file located at <project-plugin-root>/bin/project-preflight. This binary is executed with the agent's local permissions.
[DYNAMIC_EXECUTION]: The skill relies on a pre-compiled, opaque binary that is not defined in the source code. Static analysis cannot verify the internal operations, network activity, or file system interactions of this executable.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted repository metadata which could be manipulated by an attacker to influence agent behavior.
Ingestion points: Reads repo_root, integration_branch, and current_branch from the local environment (SKILL.md).
Boundary markers: None provided in the instructions to separate repository data from instructions.
Capability inventory: Executes a local binary with shell arguments (SKILL.md).
Sanitization: No evidence of validation or escaping for the repository metadata before processing.

project-preflight