The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided briefs or existing procedural text to generate or refine Agent Operating Procedures (AOPs). While this task involves handling untrusted data that will later serve as instructions for other agents, the skill provides specific structural rules and safety patterns to mitigate risks.
Ingestion points: Processes user-supplied briefs or current AOP drafts for rewriting or critiquing (SKILL.md).
Boundary markers: Enforces a strict three-section markdown structure (# GOAL, # STEPS, # NOTES) to compartmentalize procedural instructions (references/aop-structure.md).
Capability inventory: The skill itself does not invoke tools, execute scripts, or perform network operations; it is an instructional markdown skill (SKILL.md).
Sanitization: Emphasizes the use of concrete thresholds and imperative business logic, which helps minimize ambiguity and clarify agent intent.
[EXTERNAL_DOWNLOADS]: The skill mentions external resources that are verified as belonging to the official vendor or established service registries.
References the @duvoai/cli Node.js package on the NPM registry as a resource for users (SKILL.md).
Links to the vendor's official domain and documentation at duvo.ai and docs.duvo.ai (SKILL.md).
References the vendor's community skill repository on GitHub at github.com/duvoai/skills (SKILL.md).
[NO_CODE]: The skill consists entirely of markdown instructions and reference documentation without any accompanying executable scripts or binaries, significantly reducing the automated attack surface.

aop-writer