The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes project documentation (ADRs, glossary, leaf docs) as source material for generating human-facing docs. This represents a potential ingestion surface for indirect prompt injection if the source repository contains malicious instructions. However, the skill implements a strict 'derivation contract' and uses one-way 'bridge links' to limit the influence of ingested data on the agent's behavior.
Ingestion points: Processes local repository files such as ADRs, glossary, seam table, and leaf docs (referenced in SKILL.md and reference/routing.md).
Boundary markers: Uses mandatory Derived: and Sources: headers in generated files to clearly label derived content (defined in reference/templates.md).
Capability inventory: The skill is restricted to reading documentation and writing to the docs/human/ directory; it explicitly forbids code parsing and does not include network operations or arbitrary command execution in its workflow.
Sanitization: Relies on the unslopify skill for prose quality and performs a final audit before publishing artifacts.
[DYNAMIC_EXECUTION]: The skill has a hard dependency on the unslopify skill, which is resolved and loaded by 'skill identity' at runtime. This modular approach is an expected design pattern for the agent platform and relies on the platform's skill resolution mechanisms.
[EXTERNAL_DOWNLOADS]: The INSTALL.md file provides commands (npx skills add, git clone) for the user to download the skill and its dependencies from the RuralNative/RuralNative-SKILLS GitHub repository. These are standard installation procedures provided for the user and are not executed automatically by the skill during its normal operation.

document-for-humans