The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is instructed to access sensitive authentication data located at ~/.build-host/credentials.json. This access is necessary for the skill to interact with the build.host API as intended by its primary purpose.
[COMMAND_EXECUTION]: The agent is directed to perform powerful operations such as building Docker containers and routing network ports. These actions are triggered based on the contents of user-provided project directories.
[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection because it ingests and processes untrusted data from git repositories and Dockerfiles.
Ingestion points: Reads external project files, git repositories, and Dockerfiles (SKILL.md).
Boundary markers: Absent; the instructions do not specify the use of delimiters or ignore-instructions warnings for untrusted content.
Capability inventory: The agent can make network API requests to build.host and execute build/deployment tasks.
Sanitization: Absent; there are no requirements for validating or escaping the repository content before the agent processes it.

deploy-dockerfile