The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements an 'instruction override' mechanism that reads rules from a file in the project repository (.ai-factory/skill-context/aif-ci/SKILL.md) and treats them as higher-priority overrides for the agent's behavior. This architectural choice presents an indirect prompt injection surface where a malicious repository could manipulate the agent's logic for CI generation.
Ingestion points: Project-level instruction file at .ai-factory/skill-context/aif-ci/SKILL.md.
Boundary markers: The skill does not employ boundary markers or specific 'ignore' instructions to encapsulate the external rules.
Capability inventory: The skill uses Write and Bash tools, which could be misused by injected instructions to create backdoored CI workflows or execute unauthorized git commands.
Sanitization: No sanitization or validation logic is defined for the rules ingested from the repository.
[EXTERNAL_DOWNLOADS]: Fetches the Composer installer from the official getcomposer.org domain and installs the govulncheck utility from golang.org within the generated CI templates to support dependency auditing and vulnerability scanning.

aif-ci