The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to validate the generated configuration file and to perform standard container operations.
Evidence: cargo run -p chefer-cli -- check <appcipe.yml> for validation, and docker build, docker pull, docker save for image management.
[CREDENTIALS_UNSAFE]: The documentation describes a method for handling private registry authentication by setting an environment variable to a plain text username and password string.
Evidence: Mention of CHEFER_REGISTRY_AUTH=user:pass in the image source section, noting that external credential helpers are not supported.
[PROMPT_INJECTION]: The skill defines a process for generating configuration files based on user-supplied parameters, which creates an attack surface for indirect prompt injection if user inputs are not properly handled.
Ingestion points: User-provided specifications for container images, service names, and environment variables.
Boundary markers: No specific delimiters or warnings are provided to separate user-supplied data from the YAML file structure.
Capability inventory: The skill uses cargo for command execution on the host and allows defining arbitrary commands in the healthcheck field of the resulting container.
Sanitization: The instructions do not include guidance for escaping or validating user input before it is written to the configuration file.
[EXTERNAL_DOWNLOADS]: The skill describes functionality for pulling container images from external registries.
Evidence: Documentation for docker pull and registry references (e.g., redis:7.2-alpine) within the appcipe.yml schema.

write-appcipe