The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill uses dynamic context injection in SKILL.md to execute shell commands (cat and find) during the skill loading process. These commands are used to provide the agent with immediate context regarding the local .infrahub.yml configuration and the existing Python generator files within the repository.
[INDIRECT_PROMPT_INJECTION]: The skill processes external design data from Infrahub through GraphQL queries to drive infrastructure automation, creating a surface for potential indirect prompt injection.
Ingestion points: GraphQL query results defined in .gql files (e.g., queries/topology/pop.gql, queries/segment/segment.gql).
Boundary markers: The skill does not explicitly define delimiters for external data within the prompts, but it focuses on technical processing using the Infrahub SDK.
Capability inventory: The skill can execute shell commands via Bash, write and edit files, and perform network operations through the SDK client.
Sanitization: A clean_data utility is provided in rules/patterns-common.md to normalize GraphQL response structures, but no specific security-focused sanitization of the data content is mentioned.
[DYNAMIC_EXECUTION]: The skill facilitates the development and local execution of Python-based generators. It provides detailed instructions and examples for creating InfrahubGenerator subclasses and executing them using the infrahubctl generator command against a live or test environment.

infrahub-managing-generators