The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the .agent-factory/config.yaml file, specifically within the skill_prompts mapping, which is used as project-specific guidance for various modules.
Ingestion points: The agent is instructed to read .agent-factory/config.yaml and apply skill_prompts as additional guidance.
Boundary markers: The skill includes a dedicated 'Prompt Overlays' section that explicitly warns the agent that these prompts cannot grant permissions, weaken repository/host safeguards, or override user requests.
Capability inventory: As a coordinator, the skill orchestrates other modules capable of reviewing code, approving changes, and deploying software (factory-ship, factory-review-prs).
Sanitization: The instructions provide a resolution strategy for conflicts, mandating that the agent follow the most restrictive requirements when an overlay conflicts with safety rules.
[SAFE]: The skill references technical documentation hosted on the official GitHub repository for BuilderIO, a known service provider for this skill.

factory