The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process externally provided draft documents, such as business plans and pitch decks. This creates a theoretical surface for indirect prompt injection where instructions could be hidden within the draft data. However, the risk is extremely low because the skill does not possess capabilities to interact with the file system, execute shell commands, or perform network exfiltration.
Ingestion points: The skill requires "The draft output to be shipped" as input in SKILL.md.
Boundary markers: The instructions do not define explicit structural delimiters to separate user data from agent instructions.
Capability inventory: There are no scripts, tool invocations, or external process calls associated with this skill.
Sanitization: The skill focuses on manual verification and quality checks rather than automated input sanitization.
[SAFE]: The skill contains a reference to a documentation resource hosted on a well-known code hosting platform.
The URL https://github.com/peterbamuhigire/chwezi-dev-engine/blob/main/references/ai-slop-responsible-publishing-standard-2026-09-11.md corresponds to a vendor-owned resource that aligns with the skill author's identity.

anti-ai-slop